article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the phase space of qcd in the temperature and quark chemical potential @xmath1 plane is shown schematically in fig .
many of the transition lines and phases displayed in the figure are theoretical predictions with little experimental verification .
we have studied the transition as the chemical potential is varied ( dashed arrow in fig .
[ fig1 ] ) at a fixed temperature @xmath2 that is most likely above the region where a diquark condenses .
we wanted to understand the fate of the topological susceptibility @xmath3 across the transition and to decide whether its possible change occurs at the same value of @xmath1 where the vector chiral symmetry is restored and the polyakov loop signal rises @xcite .
it is known that at the finite temperature transition the signal of @xmath3 drops abruptly for yang
mills theory @xcite as well as for qcd with several values of the flavour number @xmath4 @xcite .
the present is the first work where a similar study is performed at the finite density transition ( the full account can be found in ref .
@xcite ) .
we have performed the study on the lattice by simulating the theory with two colours in order to avoid the sign problem that in the three colour theory makes the importance sampling method impracticable ( for methods to overcome this problem see @xcite and references therein ) .
we expect that this modification of true qcd has little impact on the results concerning gluon properties and in particular instanton physics .
the main differences between two and three colour qcd regard aspects of the theory that we have not studied , like the hadron spectrum and the diquark condensed phase @xcite .
in fact in the two colour theory mesons are degenerate with baryons ( which are made of two quarks ) and the diquark condensate phase becomes superfluid instead of superconducting because two quarks in two colour qcd can create a colour blind condensate .
we have simulated the model on a @xmath5 lattice at an inverse bare lattice gauge coupling @xmath6 and quark mass @xmath7 in units of lattice spacing @xmath8 .
the wilson action was used for gauge fields and the standard action for 8 flavours of quarks in the staggered formulation .
the hybrid molecular dynamics algorithm was chosen to update configurations .
they we - re separated typically by 50100 steps of the algorithm in order to well decorrelate the topology @xcite .
we followed the behaviour of several observables across the transition shown in fig .
[ fig1 ] : chiral condensate @xmath9 , polyakov loop or wilson line @xmath10 , topological susceptibility @xmath3 , baryonic density @xmath11 and average plaquette @xmath12 the measurement of the topological susceptibility requires a careful treatment of composite operators .
this quantity is defined as @xmath13 where @xmath14 is the topological charge density operator that appears in the r.h.s . of the axial singlet anomaly .
@xmath3 is related to the @xmath15 mass through the witten
veneziano mechanism @xcite and to the chiral condensate in the massless limit @xcite .
eq.([chi_continuum ] ) can be rewritten in a more compact form , @xmath16 where the total topological charge @xmath17 is defined as @xmath18 and @xmath19 is the spacetime volume .
this expression has only a formal meaning because it must be supplemented with a multiplicative renormalization for each power of the topological charge operator @xmath17 @xcite as well as a careful treatment of the contact divergences that appear when the two operators are evaluated at the same spacetime point @xcite .
a detailed description of the procedure that we have followed to extract @xmath3 can be found in @xcite . in a nutshell
, one has to define a lattice regularization @xmath20 of the topological charge density operator .
the corresponding total lattice topological charge is @xmath21 and the lattice topological susceptibility is @xmath22 .
this quantity is related to the physical susceptibility @xmath3 by the general expression @xcite @xmath23 where @xmath24 is a multiplicative renormalization of @xmath25 , @xmath8 is the lattice spacing and @xmath26 is an additive renormalization constant . in presence of composite operators
the usual renormalization program in a renormalizable theory is not enough to fully renormalize a green s function .
this is the origin of @xmath24 ( even in the case of the pure gauge theory where the topological charge operator has no anomalous dimensions ) . on the other hand part of the contact divergences that appear in the product of the two topological charge operators
must be subtracted .
this subtraction is defined as @xmath27 , i.e. @xmath26 is the value of @xmath28 in the sector of zero topological charge .
this condition guarantees the obvious requirement that @xmath3 vanishes in that sector .
one has to calculate the two constants @xmath24 and @xmath26 and insert them into eq.([chil ] ) in order to extract @xmath3 .
we shall present the results in the form of the ratio @xmath29 as a function of @xmath1 .
therefore we need not calculate @xmath24 and only @xmath26 is required .
this additive renormalization constant was computed by using a non perturbative method @xcite .
it is enough to calculate it for one single value of the chemical potential because @xmath26 is independent of infrared properties .
in fig . [ fig2 ] we show the resulting behaviour of the topological susceptibility as a function of the chemical potential .
a clear cut drop is seen at the position @xmath30 .
this sudden fall is an indication of the effective restoration of the axial singlet symmetry @xcite .
we do not show the analogous figures for the polyakov loop and chiral condensate behaviours ( see ref .
@xcite ) . instead in fig .
[ fig3 ] the derivatives of the three quantities ( topological susceptibility , polyakov loop and chiral condensate ) are superimposed to give evidence of the coincidence of the three transitions : they all happen at the same ( pseudo)critical value @xmath30 .
single points are the result of the discrete derivatives computed from the measured data while the lines are the derivatives of the interpolations made with a natural cubic spline on each data set . since the theory deconfines at the same @xmath1 where @xmath31 vanishes , the path in the @xmath2@xmath1 diagram followed in our study possibly lies above the superfluid phase ( as indicated in fig . [ fig1 ] )
@xcite . in order to give an estimate of the main results in physical units ,
we have done a second simulation of the model at zero chemical potential by varying the inverse bare lattice coupling @xmath32 on the same lattice volume and have measured the polyakov loop and chiral condensate .
the data clearly reflected the existence of a critical @xmath33 .
it was then assumed that the corresponding transition temperature @xmath34 lies in the range 100200 mev .
hence a value for the lattice spacing at @xmath35 was obtained .
this value was then run , by making use of the two loop beta function , from @xmath35 to our working @xmath36 thus obtaining the lattice spacing at this beta @xmath37 fm ( errors derived from the imprecision on @xmath35 and from the inaccuracy of the estimate for @xmath34 respectively ) .
this result allowed us to assign physical units to the various parameters of our simulation : @xmath38 mev ( errors are respectively due to the estimate of @xmath39 , the imprecision on @xmath35 and the rough approximation of @xmath34 ) and @xmath40 mev ( error on @xmath35 and on @xmath34 respectively ) . notice that the numerical value of @xmath0 is roughly compatible with half of the mass of the lightest baryon ( in the two colour theory it is made of two quarks and degenerated with the pion ) . however we take the above numbers with caution because of the large systematic errors derived from the small and coarse lattice volume , the ( wrong ) gauge group and the inexact updating algorithm . by repeating the simulation at very high values of the chemical potential , in principle one could test the perturbative region of the phase diagram .
however the advent of the so called pauli blocking makes that not viable .
as @xmath1 increases , more and more fermions are placed in the lattice volume . since it contains a countable number of sites , the pauli principle imposes a maximum allowed number of fermions that can be placed on them . in our case
the maximum allowed density is one baryon per site @xcite .
this maximum is attained for @xmath41 .
after that point , fermions are frozen and the theory becomes entirely quenched .
we have verified this statement by calculating the baryon density and the average plaquette .
their data are shown in fig .
@xmath11 becomes 1 and stays constant for all @xmath42 .
also the average plaquette stays constant after @xmath43 and this constant coincides with the value obtained from a separate monte carlo study where the pure gauge theory was simulated ( represented by the down triangle in fig .
[ fig4 ] ) .
we stress that pauli blocking is a lattice effect with no counterpart in the continuum .
9 e. shuryak , comments nucl .
( 1994 ) 235 .
b. alls , m. delia , a. di giacomo , phys .
b412 ( 1997 ) 119 .
b. alls , m. delia , a. di giacomo , nucl .
phys . b494 ( 1997 ) 281 , erratum : ibid b679 ( 2004 ) 397 . b. alls , m. delia , a. di giacomo , phys .
b483 ( 2000 ) 139 .
b. alls , m. delia , m. p. lombardo , nucl .
b752 ( 2006 ) 124 .
e. laermann , o. philipsen , ann .
53 ( 2003 ) 163 . s. hands , j. b. kogut , m. p. lombardo , s. e. morrison , nucl . phys .
b558 ( 1999 ) 327 . j. b. kogut , m. a. stephanov , d. toublan , j. j. m. verbaarschot , a. zhitnitsky , nucl .
b582 ( 2000 ) 477 .
b. alls , g. boyd , m. delia , a. di giacomo , e. vicari , phys .
b389 ( 1996 ) 107 .
b. alls et al . , phys .
d58 ( 1998 ) 071503 .
y. aoki et al . , phys .
d72 ( 2005 ) 114505 .
e. witten , nucl .
b156 ( 1979 ) 269 .
g. veneziano , nucl .
b159 ( 1979 ) 213 .
f. c. hansen , h. leutwyler , nucl .
b350 ( 1991 ) 201 .
m. campostrini , a. di giacomo , h. pana - gopoulos , phys .
b212 ( 1988 ) 206 .
b. alls , e. vicari , phys .
b268 ( 1991 ) 241 .
m. campostrini , a. di giacomo , h. pana - gopoulos , e. vicari , nucl .
b329 ( 1990 ) 683 .
a. di giacomo , e. vicari , phys .
b275 ( 1992 ) 419 .
b. alls , m. campostrini , a. di giacomo , y. gnd , e. vicari , phys .
d48 ( 1993 ) 2284 .
r. d. pisarski , f. wilczek , phys .
d29 ( 1984 ) 338 .
j. i. skullerud , s. hands , s. kim , hep
lat/0511001 ; d. toublan , a. r. zhitnitsky , hep ph/0503256
. h. kluberg stern , a. morel , o. napoly , b. petersson , nucl . phys .
b220 ( 1983 ) 447 .
|
the behaviour of the topological susceptibility in qcd with two colours and 8 flavours of quarks is studied at nonzero temperature on the lattice across the finite density transition .
it is shown that its signal drops at a ( pseudo)critical chemical potential @xmath0 .
the polyakov loop and the chiral condensate undergo their transitions at the same value .
pauli blocking supervenes at a value of the chemical potential larger than @xmath0 .
|
superclusters are complex structures , consisting of multiple galaxy clusters and groups connected by chains of galaxies
. they can reach sizes of @xmath6 mpc , making them the largest structures in the universe .
as such , they can be used to provide an estimate of the baryonic mass fraction @xmath7 . their frequency and topology may be used to test the veracity of large scale cosmological simulations . because they contain structures spanning a wide range of projected and local densities , superclusters are ideal sites for studying the variety of physical processes affecting galaxy evolution , including ram pressure stripping , mergers , tidal encounters , harassment , etc .
they are also sites of cluster - cluster interactions , allowing us to probe how cluster - scale interactions effect their constituent galaxies , and to probe possible differences between the dark and baryonic matter distributions .
recent deep , wide - field imaging campaigns have begun to reveal an increasing number of superclusters at @xmath8 .
these include a @xmath0 compact supercluster found in the red - sequence cluster survey @xcite , a structure at @xmath9 in the uk infrared deep sky survey ( ukidss ) deep extragalactic survey @xcite , and a large galaxy density enhancement at @xmath10 in the cosmos field @xcite . to increase the number of well - studied high - redshift structures , we have begun the observations of redshift evolution in large scale environments ( orelse ) survey , a systematic photometric and spectroscopic search for structure on scales @xmath11 mpc around 20 known clusters at @xmath12 ( * ? ? ?
* hereafter orelse i ) .
the cl 1604 supercluster at @xmath13 was the first supercluster detected as part of this survey and remains the most heavily studied structure at these redshifts .
it was initially discovered as two separate clusters in the plate - based survey of @xcite , with redshifts and preliminary velocity dispersions measured by ( * ? ? ?
* ; * ? ? ?
* hereafter p98 and p01 ) .
deeper ccd imaging @xcite revealed a total of four distinct galaxy density peaks in a contiguous area of @xmath14 . a large spectrographic survey using the low - resolution imaging spectrograph ( lris ; * ? ? ? * ) and the deep imaging multi - object spectrograph ( deimos ; * ? ? ?
* ) on the keck 10-m telescopes confirmed the new cluster candidates and provided improved velocity dispersions for all four supercluster components using 230 galaxies @xcite .
the large radial depth of the supercluster ( @xmath15 mpc ) in comparison to the small imaging area ( @xmath16 mpc ) spurred a wide field imaging campaign using the large format camera ( lfc ; * ? ? ? * ) on the palomar 5-m telescope .
four additional red ( @xmath17 ) galaxy overdensities , identified as candidate clusters , were detected in the larger area , bringing to eight the total number of possible clusters in this structure @xcite .
additional multi - object spectroscopy was undertaken to confirm the new cluster candidates and improve our understanding of the supercluster structure . in this paper
we utilize our complete spectroscopic data set , with @xmath18 extragalactic redshifts , to examine the three dimensional structure of the cl1604 supercluster , improve our cluster detection algorithm , and study the cluster galaxy populations .
a total of 427 galaxies have been spectroscopically confirmed within the structure , nearly doubling the number of known members .
the only comparable structures at similar redshift are the @xmath9 ukidss supercluster with five density peaks @xcite and the @xmath0 three - cluster system described in @xcite .
these two systems each have only @xmath19 spectroscopic members .
we show that even with hundreds of members spread over the supercluster , it remains difficult to obtain consistent velocity dispersions due to the presence of overlapping structures , filaments , and changes in galaxy populations between clusters .
section 2 describes the photometric data , density mapping and cluster detection , all of which are improvements on our earlier work .
the spectroscopic data is described in 3 .
cluster velocity dispersions and the associated measurement complications , along with the overall structure of the supercluster , are discussed in 4 .
we utilize our extensive spectroscopic sample to study photometric selection of cluster members using only three filters ( @xmath20 ) , and the cluster properties .
section 5 details the various fore- and background structures , including two serendipitously discovered structures at @xmath21 and an apparent wall at @xmath22 . a brief discussion of implications for optical surveys of high - redshift galaxy clusters and spectroscopic studies of cluster galaxy populations ,
is presented in 6 . throughout the paper
we use a cosmology with @xmath23 km s@xmath24 mpc@xmath24 , @xmath25 and @xmath26 .
the imaging data are presented in @xcite and @xcite .
we provide a brief review , along with a description of improvements to the photometric calibration and modifications made to the final density map since those publications .
we have also implemented a technique to estimate the contamination of our cluster catalog by false detections , described in 2.4 .
the photometric data are a combination of two pointings taken in cousins @xmath27 and gunn @xmath28 with the carnegie observatories spectroscopic multislit and imaging camera ( cosmic ; * ? ? ?
* ) and two pointings in sloan @xmath29,@xmath30 and @xmath31 with the lfc , both on the palomar 5-m telescope .
the layout of the imaging fields is shown in figure 1 of @xcite , who also describe the lfc observations and data reduction .
the details of the cosmic observations and data reduction are described in @xcite .
since the publication of @xcite , the sloan digital sky survey ( sdss ; * ? ? ?
* ) has released data in the fields covered by these pointings .
we have used this data to improve the transformation of the cosmic @xmath27 and @xmath28 magnitudes to the sdss system and to calibrate the lfc fields .
the sdss dr5 @xcite was queried for the @xmath29 , @xmath30 and @xmath31 model magnitudes in the region covered by our imaging .
objects were cross - identified between the sdss catalog and our four separate pointings ( two cosmic and two lfc ) .
the scatter between our coordinates and those of sdss were @xmath32 in both right ascension and declination , commensurate with the expected astrometric errors .
transformations from our photometry to sdss were derived independently for each filter and pointing .
the cosmic @xmath27 and @xmath28 magnitudes were converted to the sdss system using equations of the form @xmath33 for objects detected in only one band , we assumed @xmath34 , the median color of sources in the cosmic catalog .
similarly , the lfc data are recalibrated using @xmath35 the depths of the lfc images are such that any object with normal colors will be detected in all three bands , unless it as the field edges or near a saturated star .
we therefore apply no color terms to single - band detections .
there is significant overlap between the two lfc pointings , and especially the cosmic and lfc pointings . in our previous work
, we simply trimmed the catalogs to non - overlapping areas based on visually selected coordinate limits . here , we have improved the catalog combination by selecting objects based on photometric errors . first , the two lfc pointings were compared to each other . for sources appearing in both pointings ,
we use the detection with the lowest average photometric error in the three filters to create a master lfc catalog .
the same procedure is applied to the two cosmic catalogs to generate a single cosmic catalog .
these two catalogs are then compared on the basis of the @xmath29 and @xmath30 errors , with objects selected from lfc if the average photometric error @xmath36 from lfc is less than that from cosmic plus 0.05 mag .
this scheme gives preference to data from the lfc because it covers a larger area and includes photometry in @xmath31 . for sources where cosmic @xmath29 and @xmath30 photometry is chosen ,
we incorporate @xmath31 data if it is available from lfc , or from sdss if there is no lfc @xmath31 data or the lfc @xmath31 detection is flagged as bad .
the final hybrid catalog is corrected for galactic reddening using the dust maps from @xcite on an object - by - object basis .
the @xmath37 limiting magnitudes are 25.2 , 24.8 , and 23.3 in @xmath29 , @xmath30 and @xmath31 , respectively .
following @xcite , we produce a galaxy density map by adaptively smoothing a color - selected subset of galaxies in the cl1604 field . with the additional spectroscopy presented in @xcite and the improved photometric calibration discussed above , some modifications to the density mapping algorithm were made .
first , we used the observed colors of @xmath38 confirmed members to choose intervals in both the @xmath39 and @xmath40 colors that provided large numbers of supercluster galaxies and enhanced contrast relative to fore- and background structures .
the limiting colors and magnitudes used were @xmath41 , @xmath42 and @xmath43 .
these color cuts are delineated by the black rectangles in the color - magnitude and color - color diagrams in figure [ cmds ] .
we also examined the possibility of applying color cuts based on stellar population synthesis models .
synthetic galaxy spectra were generated following the prescription used by the red - sequence cluster survey ( rcs , * ? ? ?
* ) , with a @xcite model parameterized as a 0.1 gyr burst ending at @xmath44 followed by a @xmath45 gyr exponential decline .
this model predicts colors of @xmath46 and @xmath47 at @xmath0 ; they are plotted as the yellow bars and crosses in figure [ cmds ] .
the observed @xmath48 colors are about 0.2mag redder . although the bc03 model colors are within our broad color cuts , the empirical colors , by construction , provide optimal detection of structure at the redshift of interest .
tests of other models for the red galaxy population , including single starbursts with either instantaneous or exponentially declining bursts , showed that expected galaxy colors varied a few tenths of a magnitude between models .
therefore , we use our empirically confirmed color cuts to produce a final density map .
as discussed in orelse i , the star formation history chosen by the rcs yields model colors that , while within our broad color cuts , do not match perfectly the observed red sequences in any of our structures at @xmath49 .
this is unsurprising as the average galaxy s star formation history varies with environment , and there remain unresolved issues with stellar population synthesis models . as models improve and large surveys like orelse , rcs and pisces ( panoramic imaging and spectroscopy of cluster evolution with subaru , * ? ? ?
* ) obtain spectroscopy and map out the observed red sequence colors as a function of redshift , it will be possible to find bc models that better fit the data .
the color and magnitude cuts select only 722 objects out of an initial catalog of @xmath50 with @xmath51 .
an adaptive kernel smoothing is applied using an initial window of @xmath52 mpc and 10 arcsecond pixels .
we use a smaller kernel than the @xmath53 mpc radius applied in @xcite because our extensive spectroscopy showed that small groups were being blended into single detections in the density map .
the smaller kernel also enhances the contrast of small groups against the background , making detection of such low - mass systems easier .
the new density maps are shown in fig .
[ akmap ] , with the detected overdensities marked .
comparison to figure 1 of @xcite shows that similar structures are found with the new color cuts .
the density maps are used primarily to provide a visual locator for intermediate density large scale structure in the imaging field .
filaments and cluster infall regions can cover significant portions of the observed area , but at relatively low contrast , making it difficult to detect them and define their boundaries .
identification of such overdense regions is used primarily to direct placement of follow - up observations , especially slitmasks for multi - object spectroscopy .
however , it is necessary to provide a consistent detection mechanism for finding clusters and groups in the orelse survey fields , described below .
cluster detection is performed by running sextractor @xcite on the adaptively smoothed red galaxy density map . in @xcite
we relied on visual inspection of the density map and comparison to the poorest ( lowest velocity dispersion ) cluster in the field , cl1604 + 4316 ( cluster c ) . with the addition of the spectroscopy presented in @xcite and
herein , the projected overdensities can be better characterized , and detection parameters tuned to detect real structures .
we tested various combinations of detection threshold and minimum area , all taken from cluster c. this cluster still has the lowest reliably measured velocity dispersion , as detailed in 4 .
sextractor was run with these parameters , and the resulting detections compared to our spectroscopic cluster catalog to ensure that all confirmed clusters and groups were detected .
we also examined the false detection rate , described in the following section .
the final set of detection parameters has high detection efficiency ( 100% by design ) for the confirmed structures while yielding less than one false detection on average in the survey area .
a total of 10 cluster and group candidates are detected , similar to those found in @xcite .
where possible , we use the same naming and lettering convention as our previous work to refer to specific clusters .
table [ candinfo ] provides the details of each cluster .
column 1 lists the single letter denoting each cluster ( following the nomenclature of * ? ? ?
* ) , while column 2 provides the full name .
columns 3 and 4 give the updated cluster coordinates .
the remaining columns provide information on membership , redshifts , and velocity dispersions , as detailed in 4 . because one of our goals is the detection of low - mass group - like structures in the cl1604 field , our detection thresholds are generous in the sense that we may suffer from a high false detection rate , even with the stringent color cuts .
our method for setting thresholds and estimating contamination rates is detailed in orelse i ; however we outline it briefly here since the cl1604 data have been used to tune the parameters .
we use the noao deep wide field survey @xcite to study possible contamination by field galaxies having colors meeting our selection criteria .
we can not rely on our imaging survey because it is entirely targeted at known high - density regions of the universe .
thus , we use the noao dwfs to derive the statistical properties of the red galaxy distribution . the third data release ( dr3 ) in the bootes field
covers an area of nine square degrees , in the b@xmath54 , r and i filters . in brief
, we transform the dwfs @xmath27 and @xmath55 data to the sdss @xmath29 and @xmath30 systems , and apply our @xmath39 color cut and @xmath30 magnitude limit . because there is no @xmath31 photometry in the dwfs
, we then assume that field galaxies meeting only the @xmath39 cut ( numbering @xmath56 ) have a similar projected distribution as those that also meet our @xmath40 cuts , only with a higher space density .
we tested this assumption by examining the galaxy distributions far from cluster centers in density maps using our own data , applying only the @xmath39 cut and cuts in both colors .
we find no significant differences .
following @xcite and @xcite , we use the color- and magnitude - selected subset of the dwfs data to compute raleigh - levy ( rl ) parameters for galaxies meeting our color cuts , which we then use to generate simulated galaxy distributions .
these are then used to produce density maps on which we run sextractor .
the sextractor parameters detect_thresh and min_area ( @xmath57 ) are varied , using values of the galaxy density at different radii @xmath58 from the center of cluster c in cl1604 , until a set of values is found that successfully detects the confirmed clusters in cl1604 while yielding low contamination .
we find that an area of 47 pix@xmath3 , corresponding to a radius of 0.3@xmath59 mpc , allows us to detect all of the spectroscopically confirmed cluster candidates in the real data , while minimizing the number of detections in the rl simulations , which should all be chance projections .
the median number of false detections expected in the area imaged for cl1604 is @xmath60 , compared to the ten candidates detected .
figure [ rlfig ] shows an example rl simulated map @xmath61 on a side , with the eight false candidates detected using our optimal sextractor parameters marked .
the shaded box in the bottom right covers an area equal to that imaged for cl1604 .
the spectroscopic observations consist of five separate datasets : 1 .
spectra taken using lris as part of the original @xcite survey .
additional lris spectra taken in may 2000 , covering the area imaged with cosmic . 3 .
deimos spectra covering the same area taken in 2003 .
deimos spectra taken in 2005 and 2006 , using color selection and spanning a larger area imaged with lfc .
deimos spectra covering the same area as before , but including x - ray , radio , hst acs and new color - selected sources .
each sample is described in more detail below , with the layout of all masks shown in fig .
[ masks ] .
the original observations of cl1604 + 4304 and cl1604 + 4321 ( clusters a and d ) were taken in the late 1990s with lris on the keck i and ii telescopes and are detailed in @xcite , @xcite and @xcite .
these initial slitmasks targeted _ all _ galaxies with @xmath62 based on earlier imaging with lris .
it is important to note that no color selection was utilized in selecting galaxies to observe ; therefore , blue galaxies at the supercluster redshift are more likely to be observed compared to the deimos data discussed below .
redshifts were measured for 103 and 135 galaxies in the fields of cl1604 + 4304 and cl 1604 + 4321 , respectively .
imaging of a larger region with cosmic on the palomar 200-inch telescope was used to design additional lris slitmasks , targeting an area including clusters b and c. objects with magnitudes down to @xmath63 were included , with priority given to red galaxies .
six additional lris masks with 156 targets were observed in may 2000 .
median velocity errors for the lris data are 150 km s@xmath24 .
starting in may 2003 , cl1604 was observed with deimos on the keck ii telescope .
we briefly describe these observations here ; the details will be provided in a future paper releasing multi - wavelength data , including redshifts , in the cl1604 field .
the first deimos observations also spanned the region imaged with cosmic , and are detailed in @xcite .
galaxies as faint as @xmath64 were observed , with higher priority for red galaxies based on cosmic @xmath65 ; only 60% of the targeted objects met the red galaxy color cut .
we used the 1200 l / mm grating , blazed at 7500 , and @xmath66 slits , resulting in a pixel scale of 0.33 pix@xmath24 , a resolution of @xmath671.7 ( 68 km s@xmath24 ) , and typical spectral coverage from 6385 to 9015 .
data were reduced using the deep2 version of the _ spec2d _ and _ spec1d _ data reduction pipelines @xcite . since the publication of @xcite ,
the deep2 team has made available their redshift measurement pipeline _ zspec _ @xcite and all deimos redshifts have now been obtained using this package .
typical redshift errors using this method are 25 km s@xmath24 , nearly a factor of four improvement over our original measures . using the multiband lfc photometry from @xcite ,
galaxies with @xmath68 were targeted over a larger area based on both color and magnitude .
five deimos masks , labeled ce1 , fg1 , fg2 and ghf1 and ghf2 in fig .
[ masks ] were observed in 2005 and 2006 using these ground - based color selection criteria .
the cl 1604 supercluster was also the subject of a two - band 17-pointing mosaic with the advanced camera for surveys ( acs ) , a two - pointing acis - i observation with chandra @xcite , and a very large array b - array 1.4 ghz radio map @xcite .
these multi - band data were used to design additional deimos masks .
the primary sample included x - ray and radio sources within the acs mosaic , and red - sequence galaxies selected on the basis of the high - precision acs colors .
secondary samples included x - ray and radio sources outside the acs mosaic and bluer galaxies within the acs mosaic .
lower priority samples include red galaxies outside the acs mosaic , then bluer ( by as much as 0.3@xmath69 ) galaxies outside the acs area , and finally any galaxy not already selected by the above criteria .
we observed three of these masks in june 2007 .
figure [ masks ] shows the positions of the lris masks observed in may 2000 as well as all eleven deimos masks .
the locations of the cluster and group candidates are also marked .
this figure makes evident the uneven spectroscopic coverage of the supercluster .
especially noteworthy is the extensive lris coverage of clusters a and d ; because no color selection was used for the original lris masks , the blue ( presumably star - forming ) populations in these clusters will be better sampled .
nevertheless , even in regions where there is only deimos coverage , @xmath70 of the targeted galaxies were outside the red sequence .
we explore the dependence of cluster velocity dispersions on the population sampling in 4.1.3 .
in addition to the targeted objects , numerous serendipitous spectra were found by visual inspection .
these objects had their extraction windows set manually , but were otherwise run through the same pipelines as the target objects .
two of us ( rg and bl ) systematically inspected all of the _ spec2d _ created serendip one - dimensional spectra to determine whether these extractions contained genuine serendips and to confirm their redshifts .
a significant number of serendips ( 180 extragalactic objects , 138 with q@xmath71 ) were found in our deimos data .
many of these at the supercluster redshift show strong [ oii ] emission and little continuum .
these objects represent a different population than the primary targets ( red sequence galaxies ) and may impact the velocity dispersion determination , explored in 4.1.3 .
all of the spectroscopic data were combined into a single master catalog . for objects observed with both lris and deimos
, we used the deimos results due to the improved redshift accuracy with the higher dispersion grism and typically higher signal - to - noise ratio .
serendipitous spectra were matched to photometric objects from acs and/or palomar imaging .
in a few cases no counterpart was found , usually for single emission line spectra , some of which are likely to be ly@xmath72 emitters at @xmath73 @xcite .
each photometric identification was examined visually ( by rg ) to check for blends , interacting galaxies and misidentifications .
a set of flags was added to the catalog , identifying whether or not the object was detected and/or blended in either or both the acs and palomar images .
the final catalog contains a total of 1,671 unique objects . of these ,
1,215 are extragalactic objects , of which 1,138 have q=3 or q=4 ; 1,148 ( 1,089 with q=3 or 4 ) of the above are matched to a cleanly detected ( not blended ) photometric object in either the acs or ground - based imaging ; 140 are stars ; 427 are in the redshift range of the supercluster ( @xmath74 ) , of which 413 have q=3 or q=4 ; 417 ( 404 with q=3 or 4 ) of the supercluster members have clean photometry .
the top panel of figure [ spechist ] shows the redshift distribution of the q=3 or 4 extragalactic objects ( solid line ) , as well as for objects with q@xmath71 ( dotted line ) .
the bottom panel shows only the redshift range of the supercluster ( @xmath75 ) with redshift bins of @xmath76 .
in addition to a number of clear redshift peaks in the supercluster , many intervening structures are noticeable , discussed in detail in 5 .
we use only the most reliable ( q=3 or 4 ) redshifts for all of the analyses presented here .
nine of the ten cluster candidates listed in table [ candinfo ] have been extensively mapped in our spectroscopic campaign , with only cluster j lacking spectroscopic coverage .
we use our dataset to examine the redshift distribution around each cluster candidate , both to verify the structures and measure velocity dispersions . because of our generous detection threshold , we expect to find structures consistent with modest sized groups as well as richer clusters .
in some cases , clusters overlap each other even with the smallest radius used , and almost all clusters have companions within 1 @xmath77 mpc , as seen in fig .
[ akmap ] .
the complex supercluster structure makes separation of individual cluster components difficult and in some cases impossible .
information on both position and velocity must be incorporated ; even so , individual clusters or groups may exhibit significant substructure @xcite .
we use the cluster centers determined by sextractor from the density map when measuring cluster redshifts and velocity dispersions .
we note that the spectroscopic coverage varies greatly from cluster to cluster , as seen in fig .
[ masks ] .
we first construct redshift histograms within 0.5 , 1.0 and 1.5 @xmath77 mpc projected radii centered on each cluster .
these are plotted in figure [ velhists ] for clusters a - i . the shaded region , solid line and dashed line show the redshift distributions within 0.5 , 1.0 and 1.5 @xmath77 mpc , respectively . an initial redshift range ( typically spanning @xmath78 km s@xmath24 ) for each cluster is then selected by visual inspection of these plots , attempting to avoid clear redshift peaks from adjacent structures . the extensive overlap between structures results in contamination of the redshift distributions .
a good example is cluster c ( @xmath79 ) , where we see numerous galaxies at @xmath80 within a projected @xmath81 mpc radius . in this case , the redshift separation is large enough to easily distinguish the two populations .
more complicated overlaps , such as between d and f , where the redshift separation is only @xmath82 , are difficult to resolve . in such cases , neglecting the overlap may lead to overestimation of the velocity dispersion .
conversely , applying strict redshift limits to avoid such overlaps can artificially lower the measured dispersion . for each cluster
we follow the procedure described in @xcite and @xcite . after the initial redshift range is chosen , the cosmologically corrected velocities relative to the median cluster redshift
are calculated for each galaxy within the three radii .
the initial velocity windows are typically @xmath83 km s@xmath24 , sufficiently broad to avoid biasing the dispersion estimates to lower values .
these distributions are iteratively clipped at @xmath84 , where @xmath85 is the biweight dispersion computed by rostat @xcite .
typically from 0 - 2 galaxies are excluded from each cluster by this clipping .
the final dispersions are computed using the biweight estimate of the scale , with errors taken from the jackknife confidence interval on the biweight scale .
these have been shown to be well - behaved under most circumstances for modest sample sizes @xcite .
contamination from fore / background clusters , groups and filaments could bias the measured velocity dispersions and their associated errors when using rostat @xcite .
we have used quite stringent spatial and redshift cuts to establish the starting ranges for velocity dispersion measurement , which reduces the contamination .
we also examined the agreement between the @xmath84-clipped standard dispersion , the biweight estimate of the scale , and the gapped estimator , noting how many ( out of 3 maximum ) of these estimators agree within the quoted errors .
the results are presented in table [ candinfo ] .
columns 5 , 8 and 11 give the number of spectroscopically confirmed members within 0.5 , 1.0 and 1.5 @xmath77 mpc projected radii , respectively .
these numbers include only those galaxies used to compute the velocity dispersions within each radius .
columns 6 , 9 and 12 provide the median redshift for each cluster within each of the three radii , while columns 7 , 10 and 13 list the computed velocity dispersions and their errors .
columns 7 , 10 and 13 also include , in parentheses , the number of different dispersion estimators that agree within the errors . despite having fewer galaxies
, we would prefer the final velocity dispersions to be those measured within 0.5 @xmath77 mpc , since the smaller radius reduces contamination from overlapping clusters . however , only clusters a - d have sufficient members within this small radius to compute dispersions , so we quote the dispersions using the 1.0 @xmath77 mpc radius throughout .
the final redshift intervals used for this radius are shown in fig .
[ velhists ] as the vertical lines near the top of each panel .
the sole exception is cluster e , which appears to be a superposition of components related to clusters b and c. we , therefore , measure no velocity dispersion and do not show its color - magnitude diagrams .
if we assume that the velocity dispersion errors quoted in table [ candinfo ] are due purely to sample size , then we would expect the errors to decrease with the square root of the number of galaxies sampled .
this is almost exactly true for clusters a and d , and the quoted dispersions using different radii all agree within @xmath86 . for cluster b ,
the three quoted dispersions agree very well , even though the estimated error remains nearly constant with sample size . for the remaining structures , which all have much sparser sampling and appear to be intrinsically poorer ,
the effects of small numbers and contamination from interlopers make reliable dispersion estimates impossible .
a good example is cluster g , where the estimated error increases despite the larger sample size when moving from 1.0 to 1.5 @xmath77 mpc projected radius .
we see from fig .
[ masks ] that clusters g and h are very close to each other .
clearly , even with extensive spectroscopy it is not always possible to definitively assign membership of individual galaxies to specific clusters or groups .
the behavior of our dispersions and associated errors shows that systematic errors remain an important contributor , likely due to galaxies that are not physically bound to each cluster found in the redshift range and spatial region used in the calculation .
this suggests that dense sampling of cluster cores may be necessary to obtain reliable velocity dispersions .
however , comparison of such data to models requires understanding of possibly cluster mass dependent biases of the observed versus true dispersions .
another possibility , discussed by @xcite , is that the dispersions and errors are not measured correctly due to interlopers from filaments .
the presence of an approximately uniform background results in artificially inflated dispersions .
this background contribution is important as long as it is at least a modest fraction of the total sample . in @xcite
they show the exaggerated case of 500 cluster members and 500 interlopers .
we utilize clusters a and d to see if such interlopers affect significantly our dispersions . one way to estimate
the number of potential interlopers is to count galaxies within the same projected radius as the final assigned members , but in a velocity range outside the final window used for the dispersion calculation .
examination of fig .
[ velhists ] shows that there are only a few galaxies outside the final velocity windows used for clusters a and d , but less than @xmath87 km s@xmath24 away .
this provides an estimate of the density of interlopers in velocity space near each cluster . as an example , consider cluster a , using the 1.0 @xmath77 mpc radius .
the final velocity window containing all members used to calculate the dispersion is a top hat with half - width @xmath88 km s@xmath24 .
we count all galaxies @xmath89 in the same 1.0 @xmath77 mpc radius , but with velocities between @xmath90 and @xmath91 .
if the background galaxy distribution is constant with velocity , @xmath89 is also the expected number of interlopers within the cluster velocity range ( @xmath92 to @xmath93 ) . for cluster
a , we find @xmath94 for r@xmath95 mpc radii .
if these are all taken to be interlopers , they correspond to possible contamination rates of 5% , 9% and 8% , respectively . for cluster
d , we find similar results with @xmath96 and contamination rates of 3% , 4% and 8% .
this analysis assumes that _ all _ galaxies with velocities between @xmath90 and @xmath91 are interlopers , thus providing an estimate of the maximum effect that they could have . to see
if the presence of such objects could effect the velocity dispersions , we perform a monte carlo simulation with clusters a and d , using all three projected radii .
we remove @xmath89 galaxies from the velocity range @xmath97 .
one galaxy is removed in each velocity bin of width @xmath98 , ensuring that the likelihood of removal is flat across the entire velocity range , corresponding to contamination by a constant background .
the velocity dispersion and associated errors are recomputed using the new sample , and this procedure is repeated 1000 times for each cluster and radius combination .
we compute the mean velocity dispersion ( using the biweight dispersion estimator from rostat ) and mean error ( jackknife of the biweight from rostat ) for each cluster and radius combination from the 1000 monte carlo runs .
we then examine ( a ) the difference between these velocity dispersions and the original estimates and ( b ) the rms scatter within the 1000 monte carlo runs .
we find that the velocity dispersions calculated after removing potential interlopers are only @xmath99 lower than the initial estimates . as an example , the original dispersions for cluster a using the three radii are 532 , 619 , and 682 km s@xmath24 .
the interloper - corrected estimates are 532 , 595 , and 647 km s@xmath24 , corresponding to reductions of 0% , 4.0% and 5.4% , respectively .
the scatter among the monte carlo runs is 10 - 50 km s@xmath24 , depending on the cluster and radius used . adding this in quadrature to the already quoted error estimates
would only increase the errors by @xmath100 .
these results demonstrate that there is little bias in our velocity dispersions due to interlopers .
in addition to the possibility of interlopers , the spectroscopic sampling is significantly variable from cluster to cluster , as described in 3 .
this is especially true for clusters a and d , which have extensive lris data where no color cuts were applied . because the measured velocity dispersions for these ( and other ) clusters in cl1604 have decreased significantly compared to our and others earlier observations ( * ? ? ?
* p98 , p01 ) , we examine how the sampling ( by instrument and by color ) affects the current results .
first , for clusters a and d , we simply excluded the lris redshifts and measured their velocity dispersions using only the deimos data .
the deimos data have significantly smaller redshift errors , and using only the deimos data makes the instrumental sampling similar for all of the clusters .
we use the 1 @xmath101 mpc radius for this test to maintain a significant number of galaxies .
cluster a has 26 galaxies with q@xmath71 deimos redshifts , yielding a velocity dispersion @xmath102 km / s , consistent with the results including lris ( 32 galaxies , @xmath103 km / s ) . for cluster d , the deimos - only dispersion is @xmath104 km / s from 29 galaxies , compared to @xmath105 km / s from 53 galaxies when lris data are included .
these values are also consistent within the errors , although the lowered dispersion when only deimos data are used may be due to the large number of lris redshifts excluded .
the lris - observed galaxies are , on average , bluer than the deimos - observed ones , and may therefore have a different dispersion .
we note , however , that nearly half the deimos targets are also outside the red sequence .
we conclude that the inclusion of lris data for some clusters does not significantly alter our results .
we further examined if the computed dispersion is different for red sequence galaxies than for the cluster population as a whole .
such segregation was already shown by @xcite , who found higher dispersions for late - type galaxies than for early - types in rich , low redshift clusters .
this effect should be even more pronounced at high redshift , where the clusters are less evolved .
furthermore , since there is a color - density relationship , using only the red galaxies should reduce the contribution of infalling groups / filaments at the cost of smaller sample size .
we applied the same color limits used for the density mapping ( @xmath41 , @xmath42 ) and a magnitude limit of @xmath106 to the spectroscopic samples in each cluster . only clusters a
, b and d have sufficient ( though still small ) numbers of red galaxies for this exercise . using the 1 @xmath101 mpc radius to maintain a significant number of galaxies , we find : + cluster a : full sample 32 galaxies , @xmath107 km / s ; color cut 18 galaxies @xmath108 km / s + cluster b : full sample 32 galaxies , @xmath109 km / s ; color cut 12 galaxies @xmath110 km / s + cluster d : full sample 53 galaxies , @xmath105 km / s ; color cut 16 galaxies @xmath111 km / s + in all three cases , the red galaxies have a lower velocity dispersion . for cluster
a the difference is @xmath112 , while for cluster d it is only @xmath113 , and less than @xmath86 for cluster b. as shown above , these changes are not the result of instrumental coverage ( deimos or lris ) , so the differences must reflect the properties of the distinct galaxy populations .
it is interesting to note that our chandra observations find cluster a to be the most x - ray luminous system in the supercluster , exhibiting a relaxed and well - established intra - cluster medium @xcite .
if the system formed at an earlier epoch than clusters b and d , it is expected that the primordial red galaxy population would have had more time to fully virialize and establish a much different dispersion than any infalling blue galaxy population . the lack of a significant difference in the velocity dispersions of blue and red galaxies in cluster b , and to a lesser extent cluster d , may be further evidence that the systems are undergoing collapse or possible merger processes .
these results again suggest that one must exercise caution both when selecting galaxies for spectroscopic followup using colors and when interpreting the resulting cluster velocity dispersions .
our sample is extremely limited , but is one of the first to have so many cluster members at redshifts near unity .
the fact that the pattern of velocity dispersion as a function of galaxy color is not the same from cluster to cluster may imply that followup of only red - sequence galaxies might yield biased or inconsistent results , especially if the red vs. blue galaxy dispersion depends on the cluster mass or formation epoch
. this could also be true if only the most luminous cluster galaxies are used since they are much more likely to have red colors .
larger samples with deep spectroscopy of clusters in various evolutionary stages will be necessary to clarify these findings .
the velocity dispersions computed for clusters a and d , using all galaxies within 0.5 or 1 @xmath77 mpc radii , are a factor of two below the initial estimates from @xcite and @xcite .
the @xmath85 values derived here would place these clusters closer to the @xmath114 relation , alleviating the x - ray under - luminosity discussed in @xcite .
this consistency makes it tempting to assume that these are the correct dispersions to use ; however , the morphological and/or color sampling is not the same from cluster to cluster ( both due to sampling differences and population variations ) , and the x - ray luminosities may also be unreliable due to point source contamination .
these results clearly require detailed comparison to simulations to see what population mix is best for producing reliable velocity dispersions that trace the gravitating halo mass .
simplistic approaches to measuring cluster dispersions , especially at an epoch when they are still accreting significant portions of their final galaxy ( and mass ) content , are not likely to be reliable except for the most massive evolved systems .
to better understand the individual cluster structures and possible contamination from neighboring clusters and filaments , we examined the spectroscopic coverage and position - velocity distributions within each cluster .
these are shown in figure [ posvel ] for all clusters except e ( which has no clear redshift peak ) and j ( which has no spectroscopic coverage ) .
there are two panels for each cluster , each covering an area of 3.2 @xmath77 mpc on a side .
the left panels show all galaxies with @xmath115 as small black dots . essentially , these are all possible spectroscopic targets in the field regardless of priority .
red galaxies used to make the density map are shown as larger red dots .
the concentrations in the centers of most clusters are evident , and in later deimos masks these were prioritized .
black squares outline the galaxies with spectroscopic redshifts .
spectroscopic targets which met the density map color criteria are almost always cluster members , highlighting the efficiency of our color cuts in selecting supercluster members .
clusters a and d have very dense coverage because they were the first to be discovered by @xcite and were the targets of initial lris spectroscopy .
the three large circles correspond to radii of 0.5 , 1.0 and 1.5 @xmath77 mpc , used to measure velocity dispersions .
the right panels for each cluster plot all galaxies in the supercluster redshift range ( @xmath116 ) as small black dots .
the final members of each cluster ( as determined using rostat within the 1.5 @xmath77 mpc radius ) are then circled .
red circles are used for galaxies with higher recession velocities than the cluster mean , while blue circles show those blueshifted relative to the cluster mean .
the circle sizes are scaled by the ratio of the galaxies cluster - centric radial velocity to the cluster s velocity dispersion , @xmath117 .
the corresponding velocity dispersion is shown in the lower right corner of each panel , along with a circle for @xmath118 .
cluster e has no redshift limits marked and no velocity dispersions measured due to contamination from b and c. examination of fig .
[ posvel ] demonstrates that even the very well - sampled clusters ( a , b and d ) , with over thirty members within a 1 @xmath77 mpc radius , show visual evidence for substructure .
cluster d in particular appears very elongated in the ne - sw direction .
cluster b shows velocity segregation , which could be interpreted as either substructure or a triaxial cluster , elongated in the radial direction and oriented at a slight angle to the line - of - sight . to quantify the amount of substructure
, we performed @xcite tests on clusters a , b and d using the spectroscopically confirmed members within a 1.0 @xmath77 mpc radius ( 32 , 32 and 53 galaxies , respectively ) .
also known as the @xmath119 or ds test @xcite , this statistic looks for velocity substructure among galaxies near each other in projection as an indicator of merging components .
the ds statistic was shown by @xcite to be the most sensitive of the five different substructure estimators they examined , and works well even for moderate samples with only @xmath120 velocities per cluster .
this test indicates no substructure in cluster a , consistent with its appearance in fig .
[ posvel ] .
cluster b , although showing two velocity subclumps ( fig
. [ velhists ] ) , has no significant substructure based on the ds test .
this may be the result of having two subclumps very well aligned along the line - of - sight , a situation to which the ds - test is not sensitive .
cluster d has the highest likelihood of substructure , with the null hypothesis ( no substructure ) rejected at the 93% confidence level .
this is consistent with the evident elongation at an angle to the line of sight and the velocity segregation between the ne and se parts of the cluster .
recent work using both optical and x - ray data @xcite has shown that many clusters and groups are quite strongly triaxial , consistent with large simulations such as those of @xcite .
additionally , the cl1604 supercluster is extremely elongated along the line - of - sight , and we expect , based on hubble volume simulations , that the clusters will be aligned with the supercluster @xcite .
further understanding of the cluster dynamics will require detailed comparison to large @xmath121-body simulations as well as modeling of the cl1604 supercluster in particular .
the presence of a strong color - magnitude relation ( cmr ) , the red sequence for early - type galaxies , has long been observed @xcite and used for cluster detection @xcite . at the redshift of the cl1604 supercluster , galaxies become too small to classify morphologically from ground - based images .
recent work has used hst imaging to examine the cmr in high - redshift clusters @xcite , and in particular cl1604 + 4304 and cl1604 + 4321 @xcite . in a forthcoming study , we will combine our extensive spectroscopic database with a new hst acs mosaic of the cl1604 supercluster to extend such work to a wider range of environments . here , we briefly discuss the ground - based cmds of the individual clusters . figure [ cmrs ] shows the @xmath48 vs. @xmath30 and @xmath122 vs. @xmath31 cmds for each of the clusters ( note that e is excluded ; see 4.1.1 ) in cl1604 using only objects determined to be photometrically clean in the ground - based lfc and cosmic imaging .
small black points indicate all galaxies in the full imaged area having redshifts consistent with supercluster membership ( @xmath116 ) .
large red dots are galaxies determined to be members of each specific cluster at radii @xmath123 mpc
. we plot open hexagons for galaxies at @xmath124 mpc , filled squares for @xmath125 mpc , and filled triangles for @xmath126 mpc from the cluster center .
first , we note that the photometric errors are significant for fainter galaxies . at @xmath127 , for instance , the typical error is 0.1@xmath69 ; added in quadrature with another band , we expect at least 0.15@xmath69 of scatter at @xmath127 from photometric uncertainties alone .
the intrinsic width of the red sequence , due to age differences at fixed galaxy mass , is only @xmath128 in clusters at @xmath1 @xcite , similar to that observed locally @xcite .
the expected slope of the red sequence , a consequence of metallicity changes with galaxy mass , is also small , of order @xmath129 @xcite .
thus , our ground based observations are not sufficiently accurate to measure these quantities . nevertheless , a few general observations are possible .
the red sequence is most prominent in clusters a and b , which also have the highest velocity dispersions , consistent with the morphology - density and color - density relations @xcite and its disappearance in low density environments by @xmath1 @xcite .
cluster d , with a velocity dispersion only 5% lower than cluster a , shows a distinctly less pronounced red sequence .
this is illustrated in figure [ colordist ] , which shows the distribution of @xmath39 colors for spectroscopically confirmed members of clusters a and d. the left panel shows galaxies with @xmath130 , the depth of the complete lris spectroscopy in a and d. at these magnitudes , red galaxies have @xmath131 , with photometric errors of @xmath132 in @xmath29 and @xmath133 in @xmath30 , or color errors of @xmath128 .
the right panel shows fainter galaxies , with @xmath134 , where the errors are much larger .
the solid histograms indicate galaxies in a , while the dotted line is for cluster d. the excess of bluer ( @xmath135 ) galaxies in cluster d is evident , especially in the fainter population .
looking at the left panel , cluster a has eleven luminous red ( @xmath136 ) galaxies , while cluster d ( cl1604 + 4321 ) , although apparently rich , has only six .
if we examine the most luminous objects , the brightest red galaxy in d has @xmath137 ; cluster a has five more luminous galaxies , extending to @xmath138 .
the lack of luminous red galaxies coupled with the presence of numerous bluer galaxies in d results in a dilution of the red - sequence , making it appear much less prominent than in cluster a. clusters a and d have magnitude - limited spectroscopy for all objects with @xmath62 from the @xcite lris fields , covering a @xmath139 area ( @xmath140 mpc ) around each cluster , so the differences in the luminous galaxy populations between a and d are certainly real and not selection effects .
the absence of luminous galaxies in d is also not an effect of simply having fewer galaxies as a result of its lower velocity dispersion ( mass ) ; cluster a has 1.8 times as many objects with @xmath141 and @xmath142 , while we would expect only @xmath100 more if we assume @xmath143 and @xmath144 .
the discrepancy only gets worse if we look at the most luminous galaxies ( @xmath145 ) , where there are five times more in a than in d. to explain this as purely a result of a s large mass , we would have to increase the dispersion of a by @xmath146 while at the same time decreasing d s dispersion by @xmath146 , resulting in a mass ratio of 4.8 , comparable to , but still lower than , the factor of five difference in luminous red galaxies .
similarly , the fainter galaxy population differences between a and d are also physical , as they are sampled by both second - epoch lris and later deimos slitmasks , with similar coverage .
our findings are consistent with the hst results of @xcite , who had fewer redshifts but more precise photometry . combined with its elongated morphology
, these data suggest that cl1604 + 4321 ( d ) is a cluster in the process of formation , and therefore has not yet had time to build up its red sequence and assemble massive red galaxies .
the remaining structures have insufficient spectral coverage or too few members with clean ground - based imaging to make conclusive statements about their galaxy populations .
however , it is clear that many galaxies in the supercluster are associated _ not _ with the individual clusters but with the connecting filaments instead .
this implies that studies of the cluster cmr at high redshift that rely on purely photometric data may overestimate the width of the red sequence due to contamination by dynamically dissociated galaxies .
estimates of the blue galaxy fraction may be similarly compromised .
studies of low - redshift superclusters have also found that @xmath147 of supercluster members are not associated with specific clusters @xcite ; this fraction will only increase at larger lookback times as the structures are less collapsed .
as noted above , many galaxies in the redshift range of the supercluster are not associated with any of the clusters / groups .
we use our entire redshift catalog of 413 q@xmath71 confirmed members to construct a three - dimensional map of the supercluster . to produce this map
, we assume that individual galaxy redshifts directly reflect the position of the galaxy within the structure .
this is not strictly correct , since the clusters within the structure can have significant velocities relative to the hubble flow , and the galaxies within the clusters also have their own peculiar velocities .
however , we have no other data to constrain the dynamics within this region . if the structure is bound , then the clusters themselves may have significant peculiar velocities within the supercluster , further complicating any dynamical analysis .
the redshift depth of the supercluster , taken at face value , implies a radial length of order @xmath148 , while the transverse size ( using clusters a and j ) is only @xmath149 .
figures [ threed1 ] , [ threed2 ] and [ threed3 ] show three views of the supercluster . because of the nearly 8:1 axis ratio
, we have compressed the radial axis by a factor of five .
we show red and blue galaxies , divided at @xmath150 as correspondingly colored spheres .
each sphere is scaled by the observed @xmath30 luminosity of the galaxy .
figure [ threed1 ] shows the face - on view of the supercluster , as observed on the sky .
clusters a , b and d can be clearly seen .
the large red galaxy population in a is evident , as is the elongation of cluster d. figure [ threed2 ] shows the radial distribution of galaxies in cl1604 as a function of declination .
the individual clusters are clear , along with the large number of intercluster galaxies .
color segregation is easily seen , although luminous red galaxies do exist far outside the cluster centers , and luminous blue galaxies are present in the lower - mass clusters and groups .
the most luminous blue galaxies are almost exclusively found in the filaments , far from the cluster cores .
there are also large voids in the overall galaxy distribution , consistent with the structures seen in large simulations .
finally , figure [ threed3 ] shows a rotated view of the supercluster , making its overall structure clearer . from these figures
it appears that cluster b at @xmath151 is relatively isolated from the rest of the supercluster .
determining whether it is actually bound to the overall structure or a more isolated foreground object will require detailed @xmath121-body simulations mimicking this supercluster .
besides the supercluster , a variety of apparent fore- and background structures can be seen in figure [ spechist ] .
we apply three different criteria to identify coherent overdensities in both redshift and in projection .
candidate structures are selected by taking each galaxy with a measured redshift as a seed , and searching for peaks both in redshift and in projection .
we look for groups of galaxies that meet at least one of the following three requirements : * a. 15 or more concordant redshifts within @xmath152 , no spatial requirement * b. 5 or more concordant redshifts within @xmath152 and within a @xmath153mpc radius * c. 5 or more concordant redshifts within @xmath152 and within a @xmath154 mpc radius the redshift interval of @xmath152 used in peak selection corresponds to a velocity width of @xmath155 km s@xmath24 , broader than the expected dispersion of all but the most massive clusters , and is identical to the criterion of @xcite . criterion a includes structures like walls or filaments which might be rejected by the latter .
criteria b and c more closely match other surveys that do spectroscopic followup in a limited spatial region around candidate clusters .
criterion a selects a total of 18 redshift peaks outside the supercluster redshift range .
criterion b selects 6 peaks , all of which are also found by a , while criterion c selects 13 peaks , 11 of which are also selected by a. although these candidate structures are chosen based on well - defined criteria , the uneven spatial coverage , survey depth , color selection , and biases such as easier identification of emission - line objects imply that this sample is neither complete nor unbiased ( in redshift or galaxy type ) .
whether or not these structures are real or artifacts of our color selection and spatial sampling requires looking in greater detail at the spatial distributions and cmrs of galaxies in each apparent redshift peak .
these are plotted in figures [ bgclus]a - e .
each row corresponds to a different peak in the redshift histogram , starting at @xmath156 .
the first column shows the redshift histogram in the region of the specific redshift peak , with dashed lines demarcating the initial redshift range @xmath152 used to select the peak with criterion a. the left panel includes a roman numeral corresponding to the structures tallied in table [ losinfo ] , followed by letters denoting which criteria are met by the peak .
we also show the number of galaxies in the peak based on each of the criteria by which it is selected .
the median redshift is included , again using all criterion a galaxies .
the second column shows the projected distribution of galaxies in the redshift peak ( within @xmath152 ) as large dots overlaid on the overall distribution of spectroscopic objects .
the locations of clusters a - i are labeled .
if there is an apparent projected overdensity of objects as defined by criteria b and/or c , we plot a circle of radius @xmath154 mpc around the structure centroid , using the mean positions of the galaxies within the apparent projected overdensity ( except for structure vii , as detailed below ) .
when there are multiple spatial overdensities in a single redshift peak meeting criteria b and/or c , we show only the one with the most members .
the third column presents the @xmath48 vs. @xmath30 color - magnitude diagram with all photometric objects as small black dots and the structure members ( based on criterion a ) as large dots .
the color and magnitude limits for galaxies used in making the density map are shown as the black rectangle in these panels ; the highest priority deimos targets have these colors but extended to @xmath157 .
the fourth column shows the @xmath48 vs. @xmath122 color - color diagram using the same symbols .
redshifts and positions are shown for all spectroscopic objects , while photometric data points are shown only for those objects with clean ground - based photometry to avoid incorrect colors .
table [ losinfo ] provides information for all twenty of these redshift peaks .
column 1 gives the roman numeral identifier of the peak .
columns 2 and 3 give the j2000.0 coordinates . for peaks selected only by criterion a
, the coordinates are the mean of all galaxies in the @xmath152 window ; because this criterion does not require a spatial concentration , the coordinates are not necessarily reflective of any true structure center . for peaks selected by criteria
b and/or c , we use the mean coordinates from the galaxies within the overdensity defined by criterion c , which is most likely to provide a meaningful position .
column 4 lists which criteria were met by each peak , while column 5 gives the median redshift using all criterion a galaxies .
columns 6 , 7 and 8 give the number of galaxies meeting criteria a , b and c , respectively .
figure [ bgclus ] shows a remarkable diversity of structures .
some are apparently chance projections or collections of galaxies at very similar redshifts but with no clear physical association .
these types of objects are likely to be selected by criterion a but not by b or c , and include structures iv , xiii , xvi and xvii .
seven out of 20 redshift peaks meeting any of our criteria , or 35% , fall into this category .
these could be sparse walls or filaments crossing our field , or simple statistical fluctuations .
as noted earlier , the uneven spatial sampling , sparse coverage in some areas , and changes in targeting priority make it difficult to model the likelihood of these being real structures without using a full cosmological simulation .
other structures are more clearly filaments or walls , and there are distinct clusters / groups as well .
we discuss simple estimates of the false detection rate in section 5.2 .
we briefly describe each of the structures selected by our three criteria , including their potential effects on the photometric detection of cl1604 components . 1 .
selected only by criterion c , there is a concentration of five galaxies within 1 @xmath77 mpc near cluster d , and only nine within this redhift peak anywhere in the field .
most of the galaxies in this peak were observed in the original lris survey of @xcite .
the colors of these six galaxies reveal no clear red sequence , so this is not likely to be a real structure .
although galaxies in this peak meet both criteria a and c , there is no unambiguous compact , physical structure .
examining only the 11 galaxies in the larger subpeak at @xmath158 highlights the small clump north of cluster a. interestingly , most of the galaxies in this peak are h@xmath159 and [ oiii ] emitters as seen in deimos spectra .
3 . there are two strong concentrations of galaxies associated with this redshift peak . in the south , there are 8 galaxies , 7 of which are within a 1 @xmath77 mpc radius , most of which have spectra from the original lris survey of @xcite .
these eight galaxies have @xmath160 . in the north , there are 6 galaxies within a 1 @xmath77 mpc radius near clusters d and f , mostly with deimos spectra , and @xmath161 . the coordinates and dispersion reported in table [ losinfo ]
are derived using only the ten galaxies in the southern clump .
selected only by criterion a , these galaxies are scattered throughout most of the field .
there is no identifiable group or cluster , despite the modest number of galaxies and the well - defined shape of the peak .
again , almost all of the galaxies in this peak are h@xmath159 and [ oiii ] emitters as seen in deimos spectra
. 5 . identified by criterion c
, there are 8 galaxies in the @xmath152 window .
the five galaxies meeting criterion c are mostly concentrated in a tight structure only @xmath162 or 240 @xmath77 kpc across .
one of the most interesting structures , this appears to be a filament running almost directly n - s in front of the supercluster .
it has a low , group - like velocity dispersion of @xmath163 km s@xmath24 , while showing a significant red sequence .
nearly 80% of the peak members have deimos redshifts ; the lack of galaxies in the western part of the mapped field is therefore not likely due to lack of coverage .
we report the median coordinates and dispersion using all 39 members in the @xmath152 window , although there is no clear center to this structure , and many subclumps that meet criteria b and/or c. the reddest members of this structure meet our color criteria for the density map , likely enhancing the detectability of cl1604 .
these galaxies are isolated in the northeast corner of the field , and the majority have redshifts from the @xcite survey and would not meet our current color criteria
. however , the lack of galaxies at the same redshift near cluster a , where there is also heavy lris coverage , suggests that this structure is spatially isolated . 8 .
a tight clump of five galaxies is located near cluster g. four out of five of these galaxies have colors falling within our red galaxy criteria for density mapping , enhancing the detectability of cluster g. the four galaxies in closest proximity to each other are within @xmath164 , or @xmath165 km s@xmath24 , suggestive of a small group .
detected by all three criteria , there is a concentration of 11 galaxies within a 1 @xmath77 mpc radius , centered near cluster d. almost all are from the @xcite survey .
these 11 galaxies span a velocity range of only @xmath166 km s@xmath24 with @xmath167 .
they yield a velocity dispersion of @xmath168 km s@xmath24 , commensurate with being a small group . again , some of these galaxies meet the red galaxy color cut , enhancing the detectability of cluster d. 10 .
a similar structure to the previous one .
the large concentration of 16 galaxies with @xmath169 near cluster h contains only objects with deimos redshifts , many of which have red colors .
they have a velocity dispersion of @xmath170 km s@xmath24 , typical of small groups .
the redshift distribution shows multiple peaks within the @xmath152 selection window , and at least two spatial concentrations in the spatial distribution .
just south of clusters b and e we see 17 galaxies with @xmath171 within a 1.6 @xmath77 mpc radius .
they have a velocity dispersion of @xmath172 km s@xmath24 , typical of small groups .
in the northwest , between clusters h and i , there are 8 galaxies in a 1.25 @xmath77 mpc radius .
this clump has a mean redshift of @xmath173 .
unsurprisingly , a majority of these galaxies have red colors that fall within our selection window , since their redshifts are quite similar to that of the supercluster .
an extremely narrow redshift peak , there may be a small group near cluster g , selected by criteria b and c. even using all galaxies in the peak gives a dispersion of only 133 km s@xmath24 . 13 . selected only by criterion a , these galaxies are scattered throughout most of the field .
there is a concentration just west of cluster d , but insufficient to meet either of the spatial criteria .
since galaxies at this redshift would have colors strongly sampled by deimos , and the region near cluster d has extensive lris spectroscopy as well , it is unlikely that there is a significant structure associated with this peak .
a possible structure near cluster d. there are 10 galaxies in this region , with @xmath174 and @xmath175 km s@xmath24 . most of the galaxies are red and seem to form a red sequence . as with the previous structure ,
the spectroscopic sampling in this area is very dense , so it is unlikely that we have missed many of the luminous red members of this structure . 15 . a small clump near cluster a. all of the galaxies in the tight clump are from the @xcite survey . 16 .
only selected by criterion a , the redshift and spatial distributions show no evidence for a group .
17 . only selected by criterion a ,
the redshift and spatial distributions show no evidence for a group , except for a possible concentration in the northern end of the field .
18 . although meeting both criteria a and c , there is no obvious single overdensity , and the redshift distribution is quite broad . 19 .
although the redshift distribution is broad , we see a very compact group north of cluster c , and this peak meets all three selection criteria .
the 9 galaxies in this clump are contained in a region of radius @xmath176 mpc .
they have @xmath177 and @xmath178 km s@xmath24 .
this structure seems to be a low mass group .
20 . a group or poor cluster at @xmath179 .
the majority ( 12 ) of the galaxies are concentrated in a @xmath180 mpc region in the northeast corner of the observed area , whose center we have marked on the plot .
the velocity dispersion from these 12 galaxies is @xmath181 km s@xmath24 .
all of these galaxies are very faint ( @xmath182 ) and are detected spectroscopically from their oii emission alone , consistent with their broad range in colors . if the velocity dispersion is reflective of the group mass , this is the highest redshift group of such low mass known . from the third column in fig [ bgclus ]
we see that many of the structures , especially at @xmath183 , contain significant numbers of galaxies with colors consistent with supercluster membership .
some of these show projected density peaks near components of the supercluster , which likely contributed to the detectability of low - mass groups and filaments in the region . narrowing the color range
slightly would not eliminate their contributions to the density map .
the cmds and redshift distributions of the structures demonstrate the difficulty of establishing the physical nature of moderate - mass , high - redshift systems .
it is extremely common practice to detect clusters with some photometric technique , and report as few as three redshifts for confirmation .
for instance , @xcite examined spectroscopy in the fields of five @xmath184 clusters in the eso imaging survey ( eis ) cluster candidate catalog , and found numerous groups in many of the fields , often with only three concordant redshifts ( such as eisj0046 - 2951 ) .
they find multiple redshift peaks in some of their fields , and rely on the alignment of galaxies in each redshift peak with the location of the candidate derived from the initial imaging to determine confirmations and assign redshifts .
they do not specify a minimum area within which these three redshifts had to be found ( with statistical significance tests relying purely on redshift distributions with no spatial information ) , nor detail the spectroscopic completeness ( especially as a function of color ) , making statistical statements impossible .
however , inspection of their figures shows that these objects were typically within a @xmath185 arcminute diameter region inside the large spectroscopic field .
similarly , @xcite performed follow - up observations of rcs @xcite clusters , and often found only 3 - 5 concordant redshifts for the candidates .
they follow the arguments of @xcite , similar to those of @xcite and @xcite , to conclude that finding just three concordant redshifts for early type galaxies in a field with 5 arcmin radius at @xmath186 results in a @xmath187 likelihood of having a true structure .
this radius corresponds to @xmath185 arcminutes at @xmath188 . to qualitatively assess the reliability of cluster confirmation using so few concordant redshifts
, we examined each of our redshift peaks which meet only criterion a , and _ not _ criteria b or c. there are 7 such peaks comprising @xmath189 of our candidates .
this provides a sample of galaxies that are in redshift peaks but not contained in any obvious spatial structure .
we could simply assume that all of these are projections , implying that requiring _ no _ spatial coincidence results in at least a @xmath190 false positive rate for cluster confirmation .
we caution that even candidate structures meeting criteria b or c may be projections ( which would imply an increase in the false positive rate ) , and that some structures that only meet criterion a may be real ( decreasing the false positive rate ) . as noted earlier ,
applying our spectroscopic selection in detail to large cosmological simulations with realistic galaxy distributions and colors is needed to definitively understand the nature of the detected structures . to compare more directly with the studies mentioned above , we must account for our higher redshift sampling . for instance , @xcite have 266 redshifts in 5 fields .
they do not provide the area coverage of their spectroscopy , so we estimate a spectroscopic area coverage of @xmath191 sq .
arcmin based on their figures , or a redshift density of @xmath192 per sq .
this compares to our 1138 redshifts in @xmath193 sq .
arcmin , a sampling rate of 3.8 per sq .
arcmin . excluding the lris redshifts , which cover small regions around clusters a and d , we have 3.0 redshifts per sq . arcmin .
our redshift sampling is indeed greater by a factor of @xmath185 .
our area coverage is also a factor of @xmath185 greater .
however , we require five times more galaxies in each peak ( 15 as opposed to 3 ) , approximately compensating for these factors . thus , our estimated 35% false positive rate may be indicative of the rate expected in @xcite .
if instead we reduce our requirement to account purely for the difference in sampling , we need only @xmath194 galaxies per peak , but contained within a region comparable to the spectroscopic fields of @xcite , about @xmath195 .
this yields 22 candidate peaks , and they are almost identical to those found from our full area and requiring 15 galaxies in the peak , suggesting that our false positive rate is reasonable .
comparison to @xcite is more difficult , since they do not provide information on the effective area covered in each field , magnitude limits , or color selection .
however , examination of their fig .
2 shows numerous cases where there is no obvious redshift peak , especially for fields that are poorly sampled .
we also consider how rapidly the number of candidates increases with area for a fixed number of galaxies per redshift peak .
our original criterion b requires 5 concordant galaxies within a @xmath153mpc radius ; the typical area covered is 5 sq .
similarly , criterion c encompasses @xmath196 sq .
arcmin on average , while the eis area is 25 sq .
we find 6 , 13 , and 22 candidates using these search areas , respectively .
visual inspection of fig .
[ bgclus ] suggests that even criterion c may yield structures that are not obviously physical associations , and sampling even larger areas and requiring fewer than five concordant redshifts can only exacerbate this problem . alternatively , the redshift peaks in our data showing minimal spatial structure ( i.e.
, meeting only criterion a ) provide a test sample to estimate the likelihood of false positives from small projected galaxy groupings . to do this ,
we examine each galaxy in each of the redshift peaks selected only by criterion a. for each galaxy , we count the number of neighbors within 0.25 , 0.5 and 2 @xmath77 mpc radii ( 31 `` , 62 '' , 4.13 at @xmath188 ) . we then count how many projected groupings of @xmath197 galaxies are present for each peak , mimicking some of the selections used in other surveys . using the smallest radius , 3 out of 6 redshift peaks
have a group of at least 3 galaxies within the required area . increasing the test radius to 0.5 @xmath77 mpc results in all but one of these redshift peaks containing a group of three or even four nearby galaxies .
for the largest test radius , every redshift peak contains at least three and as many as five distinct groups , and these groups typically have 6 - 11 members .
these results suggest that , with our sampling , requiring only 3 galaxies even in a very small area is likely to produce false structures . at the largest radius ( 2 @xmath77 mpc )
we are covering areas similar to those in @xcite and @xcite . adjusting for the lower sampling in @xcite , nearly all of our criterion - a - only peaks
would have 3 concordant redshifts within a comparable field size .
however , only one of these peaks would have 4 or more concordant redshifts .
this simply reflects the fact that sparse spectroscopic sampling over modest regions can easily produce spurious redshift peaks .
we note that this analysis includes not just red sequence galaxies , since in some cases we do not sample the rs at the redshift of the peak .
this may increase the likelihood of finding chance projections .
if we required instead a small number of concordant redshifts and that these objects were also along the red sequence , we would have many fewer candidates and eliminate many potentially spurious groups .
however , most spectroscopic follow - up studies do not impose such a requirement .
in addition , for poor groups , we might not expect to see a strong red sequence at redshifts like those studied here , so requiring a red sequence might eliminate otherwise physical structures .
our analysis also does not consider the alignment of small redshift peaks with cluster candidates pre - selected using photometric redshifts or red sequence techniques .
the likelihood of finding a redshift peak with @xmath198 galaxies at the location of a cluster candidate and with redshift comparable to the photometric estimate is certainly much lower than finding a peak at any redshift along the line of sight .
however , we have shown that in a modest field , it is easy to find peaks with @xmath198 galaxies at almost any redshift , so using such data to confirm candidate clusters is questionable .
a competing effect is low spectroscopic completeness , decreasing the likelihood of finding close projections .
if we use simple criteria of three galaxies within a velocity window of @xmath83 km s@xmath24 and a radius of 0.25 @xmath77 mpc , every redshift peak along the line - of - sight would qualify as a physical structure . while this may be true
, it is unlikely . given the complex sampling , we would have to replicate our observations on large mock galaxy catalogs to give correct likelihoods for each structure .
this problem will persist for surveys like orelse , and for large sky surveys such as pan - starrs and lsst , where spectroscopic followup of complete samples of faint , distant galaxies will be difficult .
detection of high - redshift clusters using photometric techniques can be affected by line - of - sight projections over radial distances comparable to the redshift resolution of simple color - based techniques .
@xcite show that many clusters detected in a red - sequence type survey at @xmath1 are made up of galaxies from multiple distinct cluster - size dark matter halos .
their work does not include smaller groups , of which we expect to find many in the infall regions of high - redshift clusters .
they also demonstrate that the problem becomes substantially worse at @xmath1 than it is at @xmath199 , in part because a fixed pair of filters can not sample features such as the 4000 break over such a broad redshift range .
we find many small groups in both the fore- and background of the target structure at @xmath13 . examination of model galaxy tracks from @xcite in the @xmath39 vs. @xmath40 color - color space for both galaxies with a single burst at @xmath200 and the rcs model ( a 0.1 gyr burst ending at @xmath44 with a @xmath45 gyr exponential decline thereafter ) shows that colors for galaxies over a moderately broad redshift range ( @xmath201 ) are quite similar , with the differences in color comparable to ground - based photometric errors ( orelse i ) .
therefore , any survey relying on only two- or three - color moderate - depth ground - based imaging is subject to this type of confusion .
the surveys will be biased toward detecting projected systems , the same problem that has plagued lower redshift optical cluster catalogs @xcite .
even with large cosmological simulations , the detailed effects of such projections on cluster detection , number counts , and mass estimates will be difficult to quantify .
the most recent studies @xcite using the millennium simulation and a red - sequence cluster finder are still hampered by the inability of simulations to reproduce the observed colors of cluster galaxies .
extensive spectroscopy of at least a subset of clusters found using optical and x - ray surveys is needed to understand the contamination rates , redshift distributions , substructure , and other cluster properties ; such work is only now under way @xcite .
given that such projections can only enhance the detectability of lower mass systems , as well as inflate velocity dispersions and increase the weak lensing signal , mass estimates for single bound cluster components using such techniques will always tend to be overestimated .
a striking example is cluster a ( cl1604 + 4304 ) , whose initial velocity dispersion estimate was 1200 km s@xmath24 @xcite . adding many more spectra , but not tightly restricting the area sampled
, @xcite found @xmath202 km s@xmath24 .
we now find that the velocity dispersion can be as low as @xmath203 km s@xmath24 , more than a factor of two below the initial measurement .
comparison of velocity dispersions derived using a variety of galaxy subpopulations ( as done in 4.3 ) with simulations is necessary to determine the optimal mass tracer
. suggested that structures associated with , but not bound to , a cluster can increase errors in lensing mass estimates twofold .
our data shows that such filaments may be common at @xmath1 .
furthermore , completely independent structures along the line - of - sight can be cause for concern , even at low redshift @xcite .
in cl1604 , we find moderate - mass groups at @xmath5 directly projected onto clusters at @xmath13 .
both types of structures can affect the derived mass profiles .
typically , nfw profiles @xcite are fit to clusters in both observations and simulations ; clusters have been observed with nfw concentrations _
incompatible with simulated clusters .
very recent simulations of weak lensing by clusters plus foreground group - size halos @xcite demonstrates that line - of - sight contamination by even low - mass systems can both decrease _ _ and increase the measured value of _ c _ , depending on the geometry .
it is not yet known how much this can affect the overall distribution of observed concentrations and masses from weak lensing . implementing techniques that incorporate multi - wavelength data ( x - ray , sunyaev - zeldovich , optical ) can overcome some of these issues , but requires both extensive data and computational power @xcite , limiting it to rich , low - redshift clusters or only the most massive structures at higher redshift for the time being . with somewhat less data ,
it is possible to compare x - ray and weak lensing masses for clusters .
recent work using observational data has shown moderate agreement between these two techniques , but discrepancies are plentiful and systematic errors are much too large for precision cosmological measurements @xcite .
in contrast , some lensing simulations claim that high completeness , low contamination , and good mass fidelity can be achieved with weak lensing @xcite , but this has yet to be proven with large samples of real data .
weak lensing has already been used to estimate the mass of cl1604 + 4304 ( cluster a , * ? ? ?
* ) , resulting in a mass of @xmath204 , corresponding to a velocity dispersion of @xmath205 km s@xmath24 .
this is significantly higher than the @xmath206 km s@xmath24 estimated from the cluster s x - ray temperature of @xmath207 kev @xcite .
the velocity dispersion presented here lies in between these two estimates , but is much more consistent with the x - ray temperature .
earlier work on a much smaller region using hst wfpc2 data suggested multiple mass peaks associated with this structure @xcite .
cl1604 + 4304 is the most isolated component of the supercluster , with the only foreground structures being a group at @xmath208 and a possible wall at @xmath209 .
nevertheless , its estimated mass varies by a factor of nearly two .
improved x - ray measurements from deep chandra observations @xcite and lensing data from a recent acs mosaic , combined with our spectroscopic database , may help resolve this discrepancy . instead of measuring individual cluster masses , one could avoid some of the above problems with a sufficiently large and deep survey by calculating the mass on very large ( @xmath210 mpc ) scales over several superclusters , and comparing the results to simulations and structure formation theories , but such surveys remain in the future . with excellent deep photometric data ,
it is also possible to use weak lensing tomography , incorporating photometric redshifts to map all the galaxies along the line - of - sight .
however , even the best photometric redshifts have errors which are equal to or greater than the entire redshift depth of the cl1604 structure @xcite . as a result ,
uncertainties in the bias and scatter of photo-@xmath211 estimators with redshift remain a problem @xcite , and will require extensive spectroscopy to calibrate . finally , we note that cl1604 is an optically selected large scale structure . to whatever extent
it is not representative of all structures at similar redshifts , any conclusions based on cl1604 alone may be biased . for this reason
, the orelse survey has included a variety of structures originally detected using x - ray , optical and even radio techniques .
once the survey is complete , we can address biases due to the cluster detection technique .
within cl1604 , the supercluster members were targeted based on their red colors ( see 3 ) , so that the blue galaxy population may be undersampled ; nevertheless , @xmath70 of the targets are outside the red sequence .
the extensive spectroscopic database described here is being used for a variety of projects .
detailed analysis of line strengths as a function of position in the structure , local density , and host cluster properties are being derived by stacking spectra @xcite .
the redshift range of the supercluster along with our spectral setup forces the use of the [ oii ] doublet at 3727 as a proxy for h@xmath72 emission and thus ongoing star - formation activity . since [ oii ] has been shown to be a flawed tracer @xcite , we are using near - ir spectroscopy to search for h@xmath72 emission in a subset of our confirmed supercluster members in order to determine the contribution of non - stellar ( i.e. nuclear ) sources to the observed [ oii ] emission .
we have also serendipitously detected a significant number of @xmath73 ly@xmath72 sources in the deimos data , and their properties will be described in @xcite .
a 17-pointing acs mosaic of the field is being used to examine the cmds of individual clusters , in the manner of @xcite and @xcite , who used two earlier pointings centered on clusters a and d. these images also allow us to compare galaxy colors and morphologies with their spectroscopic properties , and relate them to their location within the supercluster .
deep chandra observations are being used to map the distribution of agn within the supercluster and determine the x - ray luminosity ( and hence mass ) of individual clusters in the complex @xcite .
our spectroscopic database , along with our acs imaging , will allow us to examine the spectral and morphological properties of agn host galaxies , as well as their local environments , in order to shed light on the mechanisms which trigger nuclear activity in these systems .
deep vla 1.4ghz mapping of the field shows both bright agn - like sources and an excess of faint , possibly star - bursting galaxies @xcite .
preliminary results already suggest signs of environmental segregation , with the radio - loud population found in dense regions which the radio - faint and x - ray detected agn avoid .
deep spitzer irac and mips maps have been obtained , providing both rest - frame k - band luminosities for estimating the stellar mass content of member galaxies , and search for obscured star formation that is becoming more prevalent at higher redshifts .
ground - based k - band imaging is also being used for stellar mass estimation .
combining all of these observations will allow detailed sed fitting and modeling of the star formation histories of supercluster members . for galaxies without spectroscopy , the extensive multi - wavelength data will be used to estimate redshifts photometrically , both as input to lensing models and to better map the supercluster structure .
we have presented the results of an extensive spectroscopic campaign to map the cl 1604 supercluster at @xmath13 . with over 400 confirmed members
, this database provides coverage comparable to studies of local structures , but at a lookback time of 7 gyr .
the large number of redshifts is crucial for confirming the individual clusters , measuring accurate velocity dispersions , and finding contamination due to fore- and background clusters .
a three - dimensional map of the supercluster is constructed , making possible the comparison of such structures to large simulations .
we also report the discovery of a poor cluster at @xmath179 with at least 15 confirmed members within a @xmath212 km s@xmath24 window and @xmath213 mpc radius , making it one of the highest redshift structures selected spectroscopically
. this dataset forms the foundation for a variety of follow - up studies , using many wavelengths to identify preferred sites for inducing and quenching star formation and agn activity .
the extensive spectroscopy allows identification of samples of rare objects , including radio and x - ray sources within the supercluster . without the large number of redshifts , it is impossible to definitively associate individual objects with the supercluster due to the many sources along the line of sight that are not physically associated with the structure . as an example ,
@xcite and @xcite show that x - ray and radio sources within the supercluster are a very small fraction of the total number of such sources in the field .
the same is true for infrared sources such as starbursts .
construction of luminosity functions for these rarer sources has traditionally hampered by the lack of redshifts @xcite ; our survey is designed to alleviate these problems .
the only comparable project is the pisces program , with deep optical imaging and spectroscopy of high-@xmath211 clusters with subaru , but their targets are typically more massive clusters and their spectroscopic followup is much less complete .
field surveys such as goods , aegis and cosmos provide similar data to orelse , both in terms of multi - wavelength and spectroscopic coverage , but lack the sampling of dense environments .
for instance , cosmos contains just one large scale structure at @xmath183 @xcite .
such detailed mapping , encompassing large ( @xmath11 mpc ) areas will be essential for a statistical sample of clusters over a wide redshift baseline to understand the diversity of structures they are embedded in .
the redshift depth of the supercluster is comparable to photometric redshift errors achievable from optical ground based surveys , and especially two - band cluster surveys , potentially enhancing its detectability . at such high redshifts ,
photometric errors are larger for faint galaxies , photometric redshift errors increase , the change in color of the red sequence with unit @xmath214 is reduced ( see fig . 1 of * ? ? ?
* ) , and clusters are still in the process of merging and accreting their galaxies .
all of these effects complicate our ability to reliably construct cluster catalogs over a wide range of redshifts .
future programs relying on photometric redshifts , weak lensing , s - z , and x - ray data ( in particular , luminosities ) to estimate cluster masses must understand the biases inherent in each technique , especially when using large samples to constrain cosmological parameters . while large cosmological simulations can provide insight into these issues , their current inability to accurately reproduce galaxy properties , especially colors , means that more observations are necessary .
the orelse survey will provide such data on a variety of structures at @xmath215 , forming an invaluable resource for both galaxy evolution and cosmological studies .
this material is based upon work supported by the national aeronautics and space administration under award no .
nng05gc34zg for the long term space astrophysics program .
support for program hst - go-08560.05-a was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
the spectrographic 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 .
the observing staff , telescope operators , and instrument scientists at keck provided a great deal of assistance .
we would also like to thank the deep2 team for making their software available to us , and especially m. cooper for his assistance with implementation .
llrrcrrlrrrlrrrl a & cl1604 + 4304 & 241.08946 & 43.07613 & & 21 & 0.89838 & 532@xmath216127 ( 2 ) & & 32 & 0.89861 & 619@xmath21696 ( 3 ) & & 36 & 0.89832 & 682@xmath216109 ( 3 ) + b & cl1604 + 4314 & 241.09890 & 43.23550 & & 20 & 0.86574 & 756@xmath21683 ( 3 ) & & 32 & 0.86531 & 811@xmath21676 ( 3 ) & & 47 & 0.86569 & 767@xmath21676 ( 3 ) + c & cl1604 + 4316 & 241.03162 & 43.26313 & & 8 & 0.93451 & 439@xmath216256 ( 1 ) & & 18 & 0.93511 & 313@xmath21641 ( 3 ) & & & & + d & cl1604 + 4321 & 241.13865 & 43.35343 & & 28 & 0.92351 & 559@xmath216171 ( 2 ) & & 53 & 0.92280 & 590@xmath216112 ( 2 ) & & 69 & 0.92340 & 668@xmath216103 ( 3 ) + e & cl1604 + 4314b & 241.02815 & 43.23343 & & & & & & & & & & & & + f & cl1605 + 4322 & 241.21314 & 43.37091 & & 7 & 0.93659 & 142@xmath216297 ( 1 ) & & 9 & 0.93608 & 173@xmath216220 ( 1 ) & & 13 & 0.93532 & 435@xmath216129 ( 2 ) + g & cl1604 + 4324 & 240.93754 & 43.40520 & & 6 & 0.89932 & & & 13 & 0.90075 & 409@xmath216101 ( 3 ) & & 21 & 0.90210 & 477@xmath216152 ( 3 ) + h & cl1604 + 4322 & 240.89648 & 43.37309 & & 6 & 0.85292 & & & 10 & 0.85226 & 302@xmath21664 ( 3 ) & & 11 & 0.85243 & 312@xmath21657 ( 3 ) + i & cl1603 + 4323 & 240.79691 & 43.39176 & & 5 & 0.90230 & & & 7 & 0.90238 & 359@xmath216140 ( 3 ) & & 7 & 0.90238 & 359@xmath216140 ( 3 ) + j & cl1604 + 4331 & 240.91905 & 43.51648 & & & & & & & & & & & & [ candinfo ] lrrccrrr
i & 241.13561 & 43.37704 & c & 0.459 & & & 5 + ii & 241.05078 & 43.12914 & ac & 0.470 & 17 & & 5
+ iii & 241.09854 & 43.05963 & abc & 0.496 & 22 & 5 & 6 + iv & 241.02019 & 43.24290 & a & 0.544 & 17 & & + v & 240.98975 & 43.34832 & c & 0.568 & & & 5 + vi & 241.07777 & 43.26100 & abc & 0.599 & 39 & 5 & 9 + vii & 241.11231 & 43.38718 & abc & 0.620 & 16 & 5 & 8 + viii & 240.94653 & 43.39078 & ac & 0.642 & 15 & & 5 + ix & 241.14468 & 43.38072 & abc & 0.696 & 21 & 6 & 11 + x & 240.87280 & 43.40267 & ac & 0.730 & 32 & & 9 + xi & 241.06424 & 43.19683 & ac & 0.777 & 38 & & 9 + xii & 240.94727 & 43.39270 & abc & 0.789 & 16 & 5 & 8 + xiii & 241.08856 & 43.30186 & a & 0.809 & 15 & & + xiv & 241.09376 & 43.33907 & a & 0.822 & 21 & & + xv & 241.09265 & 43.06344 & a & 0.829 & 15 & & + xvi & 241.06491 & 43.26877 & a & 0.971 & 18 & & + xvii & 241.05082 & 43.30216 & a & 0.982 & 24 & & + xviii & 241.09734 & 43.32171 & ac & 1.032 & 22 & & 5 + xix & 241.05106 & 43.30613 & abc & 1.177 & 22 & 5 & 8 + xx & 241.13109 & 43.35544 & a & 1.207 & 15 & & + [ losinfo ]
|
the cl1604 supercluster at @xmath0 is one of a small handful of such structures discovered in the high redshift universe , and is the first target observed as part of the observations of redshift evolution in large scale environments ( orelse ) survey . to date ,
cl1604 is the largest structure mapped at @xmath1 , with the most constituent clusters and the largest number of spectroscopically confirmed member galaxies . in this paper
we present the results of a spectroscopic campaign to create a three - dimensional map of cl1604 and to understand the contamination by fore- and background large scale structures .
combining new deep imaging multi - object spectrograph observations with previous data yields high - quality redshifts for 1,138 extragalactic objects in a @xmath2deg@xmath3 region , 413 of which are supercluster members .
we examine the complex three dimensional structure of cl1604 , providing velocity dispersions for eight of the member clusters and groups .
our extensive spectroscopic dataset is used to examine potential biases in cluster velocity dispersion measurements in the presence of overlapping structures and filaments .
we also discuss other structures found along the line - of - sight , including a filament at @xmath4 and two serendipitously discovered groups at @xmath5 .
|
whereas classical statistical inference is performed in a centralized manner , many modern scientific problems and engineering systems are inherently _ decentralized _ : data are distributed , and can not be aggregated due to various forms of communication constraints .
an important example of such a decentralized system is a sensor network @xcite : a set of spatially - distributed sensors collect data about the environmental state ( e.g. , temperature , humidity or light ) .
typically , these networks are based on ad hoc deployments , in which the individual sensors are low - cost , and must operate under very severe power constraints ( e.g. , limited battery life ) . in statistical terms ,
such communication constraints imply that the individual sensors can not transmit the raw data ; rather , they must compress or quantize the data for instance , by reducing a continuous - valued observation to a single bit and can transmit only this compressed representation back to the fusion center . by now , there is a rich literature in both information theory and statistical signal processing on problems of decentralized statistical inference .
a number of researchers , dating back to the seminal paper of tenney and sandell @xcite , have studied the problem of hypothesis testing under communication - constraints ; see the survey papers @xcite and references therein for overviews of this line of work .
the hypothesis - testing problem has also been studied in the information theory community , where the analysis is asymptotic and shannon - theoretic in nature @xcite
. a parallel line of work deals with problem of decentralized estimation .
work in signal processing typically formulates it as a quantizer design problem and considers finite sample behavior @xcite ; in contrast , the information - theoretic approach is asymptotic in nature , based on rate - distortion theory @xcite . in much of the literature on decentralized statistical inference
, it is assumed that the underlying distributions are known with a specified parametric form ( e.g. , gaussian ) .
more recent work has addressed non - parametric and data - driven formulations of these problems , in which the decision - maker is simply provided samples from the unknown distribution @xcite .
for instance , nguyen et al .
@xcite established statistical consistency for non - parametric approaches to decentralized hypothesis testing based on reproducing kernel hilbert spaces .
luo @xcite analyzed a non - parametric formulation of decentralized mean estimation , in which a fixed but unknown parameter is corrupted by noise with bounded support but otherwise arbitrary distribution , and shown that decentralized approaches can achieve error rates that are order - optimal with respect to the centralized optimum .
this paper addresses a different problem in decentralized non - parametric inference namely , that of estimating an arbitrary quantile of an unknown distribution . since there exists no unbiased estimator based on a single sample ,
we consider the performance of a network of @xmath0 sensors , each of which collects a total of @xmath9 observations in a sequential manner .
our analysis treats the standard fusion - based architecture , in which each of the @xmath0 sensors transmits information to the fusion center via a communication - constrained channel . more concretely , at each observation round , each sensor is allowed to transmit a single bit to the fusion center , which in turn is permitted to send some number @xmath1 bits of feedback . for a decentralized protocol with @xmath10 bits of feedback , we prove that the algorithm achieves the order - optimal rate of the best centralized method ( i.e. , one with access to the full collection of raw data ) .
we also consider a protocol that permits only a single bit of feedback , and establish that it achieves the same rate .
this single - bit protocol is advantageous in that , with for a fixed target mean - squared error of the quantile estimate , it yields longer sensor lifetimes than either the centralized or full feedback protocols . the remainder of the paper is organized as follows .
we begin in section [ secdecent ] with background on quantile estimation , and optimal rates in the centralized setting .
we then describe two algorithms for solving the corresponding decentralized version , based on @xmath11 and @xmath12 bit of feedback respectively , and provide an asymptotic characterization of their performance .
these theoretical results are complemented with empirical simulations .
section [ sectheory ] contains the analysis of these two algorithms . in section [ secext ] ,
we consider various extensions , including the case of feedback bits @xmath13 varying between the two extremes , and the effect of noise on the feedforward link .
we conclude in section [ secdiscussion ] with a discussion .
in this section , we begin with some background material on ( centralized ) quantile estimation , before introducing our decentralized algorithms , and stating our main theoretical results .
we begin with classical background on the problem of quantile estimation ( see serfling @xcite for further details ) .
given a real - valued random variable @xmath14 , let be its cumulative distribution function ( cdf ) , which is non - decreasing and right - continuous . for any @xmath15 , the @xmath16-quantile of @xmath14 is defined as @xmath17 .
moreover , if @xmath18 is continuous at @xmath19 , then we have @xmath20 . as a particular example , for @xmath21 , the associated quantile is simply the median .
now suppose that for a fixed level @xmath22 , we wish to estimate the quantile @xmath23 .
rather than impose a particular parameterized form on @xmath18 , we work in a non - parametric setting , in which we assume only that the distribution function @xmath18 is differentiable , so that @xmath14 has the density function @xmath24 ( w.r.t lebesgue measure ) , and moreover that @xmath25 for all @xmath26 . in this
setting , a standard estimator for @xmath27 is the _ sample quantile _
@xmath28 where @xmath29 denotes the empirical distribution function based on i.i.d .
samples @xmath30 . under the conditions given above
, it can be shown @xcite that @xmath31 is strongly consistent for @xmath27 ( i.e. , @xmath32 ) , and moreover that asymptotic normality holds @xmath33 so that the asymptotic mse decreases as @xmath34 , where @xmath35 is the total number of samples .
although this @xmath36 rate is optimal , the precise form of the asymptotic variance need not be in general ; see zielinski @xcite for in - depth discussion of the optimal asymptotic variances that can be obtained with variants of this basic estimator under different conditions .
we consider the standard network architecture illustrated in figure [ fignetwork ] .
there are @xmath0 sensors , each of which has a dedicated two - way link to a fusion center .
we assume that each sensor @xmath37 collects independent samples @xmath38 of the random variable @xmath39 with distribution function we consider a sequential version of the quantile estimation problem , in which sensor @xmath40 receives measurements @xmath41 at time steps @xmath42 , and the fusion center forms an estimate @xmath43 of the quantile .
the key condition giving rise to the decentralized nature of the problem is that communication between each sensor and the central processor is constrained , so that the sensor can not simply relay its measurement @xmath38 to the central location , but rather must perform local computation , and then transmit a summary statistic to the fusion center .
more concretely , we impose the following restrictions on the protocol . first , at each time step @xmath44 , each sensor @xmath45 can transmit a single bit @xmath46 to the fusion center .
second , the fusion center can broadcast @xmath47 bits back to the sensor nodes at each time step .
we analyze two distinct protocols , depending on whether or sensors .
each sensor is permitted to transmit a @xmath12-bit message to the fusion center ; in turn , the fusion center is permitted to broadcast @xmath1 bits of feedback.,title="fig : " ] for each protocol , all sensors are initialized with some fixed @xmath48 .
the algorithms are specified in terms of a constant @xmath49 and step sizes @xmath50 that satisfy the conditions @xmath51 the first condition ensures infinite travel ( i.e. , that the sequence @xmath52 can reach @xmath27 from any starting condition ) , whereas the second condition ( which implies that @xmath53 ) is required for variance reduction .
a standard choice satisfying these conditions and the one that we assume herein is @xmath54 . with this set - up
, the @xmath11-bit scheme consists of the steps given in table [ tabmbf ] .
although the most straightforward feedback protocol is to broadcast back the @xmath55 received bits @xmath56 , as described in step ( c ) , in fact it suffices to transmit only the @xmath57 bits required to perfectly describe the binomial random variable @xmath58 in order to update @xmath43 . in either case , after the feedback step
, each sensor knows the value of the sum @xmath59 , which ( in conjunction with knowledge of @xmath55 , @xmath60 and @xmath61 ) allow it to compute the updated parameter @xmath62 . finally ,
knowledge of @xmath62 allows each sensor to then compute the local decision in the following round .
the 1-bit feedback scheme detailed in table [ tab1bf ] is similar , except that it requires broadcasting only a single bit ( @xmath63 ) , and involves an extra step size parameter @xmath64 , which is specified in the statement of theorem [ thm1bf ] .
after the feedback step of the 1-bf algorithm , each sensor has knowledge of the aggregate decision @xmath63 , which ( in conjunction with @xmath61 and the constant @xmath65 ) allow it to compute the updated parameter @xmath62 .
knowledge of this parameter suffices to compute the local decision .
we now state our main results on the convergence behavior of these two distributed protocols . in all cases , we assume the step size choice @xmath66 . given fixed @xmath22 , we use @xmath27 to denote the @xmath60-level quantile ( i.e. , such that @xmath67 ) ; note that our assumption of a strictly positive density guarantees that @xmath27 is unique .
[ thmmbf ] for any @xmath22 , consider a random sequence @xmath68 generated by the @xmath55-bit feedback protocol .
then \(a ) for all initial conditions @xmath48 , the sequence @xmath43 converges almost surely to the @xmath60-quantile @xmath27 .
\(b ) moreover , if the constant @xmath69 is chosen to satisfy @xmath70 , then @xmath71 } \
; \frac{1}{{\ensuremath{m}}}\right),\ ] ] so that the asymptotic mse is @xmath72 .
_ remarks : _
after @xmath2 steps of this decentralized protocol , a total of @xmath73 observations have been made , so that our discussion in section [ secbackground ] dictates ( see equation ) that the optimal asymptotic mse is @xmath74 .
interestingly , then , the @xmath57-bit feedback decentralized protocol is order - optimal with respect to the centralized gold standard . before stating the analogous result for the 1-bit feedback protocol , we begin by introducing some useful notation .
first , we define for any fixed @xmath75 the random variable @xmath76 note that for each fixed @xmath77 , the distribution of @xmath78 is binomial with parameters @xmath55 and @xmath79 .
it is convenient to define the function @xmath80 with domain @xmath81 \times [ 0,1]$ ] . with this notation
, we have @xmath82 again , we fix an arbitrary @xmath22 and let @xmath27 be the associated @xmath60-quantile satisfying [ thm1bf ] given a random sequence @xmath68 generated by the @xmath12-bit feedback protocol , we have 1 . for any initial condition , the sequence @xmath83 .
2 . suppose that the step size @xmath84 is chosen such that or equivalently such that @xmath85 then @xmath86}{2 \gamma_{{\ensuremath{m}}}({\ensuremath{\theta^ * } } ) - 1 } \right)\ ] ] 3 .
if we choose a _ constant step size _
@xmath87 , then as the asymptotic variance behaves as @xmath88,\ ] ] so that the asymptotic mse is @xmath89 .
4 . if we choose a _ decaying step size _
@xmath90 , then @xmath91,\ ] ] so that the asymptotic mse is @xmath92 .
it is interesting to compare the performance of each proposed decentralized algorithm to the centralized performance .
considering first the @xmath57-bf scheme , suppose that we set @xmath93 .
using the formula from theorem [ thmmbf ] , we obtain that the asymptotic variance of the @xmath55-bf scheme with this choice of @xmath94 is given by @xmath95 , thus matching the asymptotics of the centralized quantile estimator .
in fact , it can be shown that the choice @xmath96 is optimal in the sense of minimizing the asymptotic variance for our scheme , when @xmath94 is constrained by the stability criterion in theorem [ thmmbf ] . in practice , however , the value @xmath97 is typically not known , so that it may not be possible to implement exactly this scheme .
an interesting question is whether an adaptive scheme could be used to estimate @xmath97 ( and hence the optimal @xmath94 simultaneously ) , thereby achieving this optimal asymptotic variance .
we leave this question open as an interesting direction for future work .
turning now to the algorithm @xmath12-bf , if we make the substitution @xmath98 in equation , then we obtain the asymptotic variance @xmath99 } \ ; \frac{1}{{\ensuremath{m}}}.\ ] ] since the stability criterion is the same as that for @xmath6-bf , the optimal choice is @xmath100 . consequently , while the @xmath101)$ ] rate is the same as both the centralized and decentralized @xmath55-bf protocols , the pre - factor for the @xmath12-bf algorithm is @xmath102 times larger than the optimized @xmath55-bf scheme .
however , despite this loss in the pre - factor , the @xmath12-bf protocol has substantial advantages over the @xmath55-bf ; in particular , the network lifetime scales as @xmath103 compared to @xmath104 for the @xmath57-bf scheme .
we now provide some simulation results in order to illustrate the two decentralized protocols , and the agreement between theory and practice . [ cols="^,^,^ " , ] we now briefly consider the effect of communication noise on our algorithms .
there are two types of noise to consider : ( a ) _ feedforward _ , meaning noise in the link from sensor node to fusion center , and ( b ) _ feedback _ , meaning noise in the feedback link from fusion center to the sensor nodes . here
we show that feedforward noise can be handled in a relatively straightforward way in our algorithmic framework . on the other hand
, feedback noise requires a different analysis , as the different sensors may loose synchronicity in their updating procedure .
although a thorough analysis of such asynchronicity is an interesting topic for future research , we note that assuming noiseless feedback is not unreasonable , since the fusion center typically has greater transmission power .
focusing then on the case of feedforward noise , let us assume that the link between each sensor and the fusion center acts as a binary symmetric channel ( bsc ) with probability @xmath105 .
more precisely , if a bit @xmath106 is transmitted , then the received bit @xmath107 has the ( conditional ) distribution @xmath108 with this bit - flipping noise , the updates ( both equation and ) need to be modified so as to correct for the bias introduced by the channel noise . if @xmath60 denotes the desired quantile , then in the presence of bsc(@xmath109 noise , both algorithms should be run with the modified parameter @xmath110 note that @xmath111 ranges between @xmath60 ( for the noiseless case @xmath112 ) , to a quantity arbitrarily close to @xmath113 , as the channel approaches the extreme of pure noise ( @xmath114 ) .
the following lemma shows that for all @xmath115 , this adjustment suffices to correct the algorithm .
moreover , it specifies how the resulting asymptotic variance depends on the noise parameter : [ propnoise ] suppose that each of the @xmath55 feedforward links from sensor to fusion center are modeled as i.i.d .
bsc channels with probability @xmath116
. then the @xmath55-bf or @xmath12-bf algorithms , with the adjusted @xmath111 , are strongly consistent in computing the @xmath60-quantile . moreover , with appropriate step size choices , their asymptotic mses scale as @xmath117 with respective pre - factors given by [ eqndefnvnoise ] @xmath118 } \\
\label{eqndefnvnoise1 } v_1({\ensuremath{\epsilon } } ) & { \ensuremath { : \ , = } } & \left [ \frac{{\ensuremath{k}}^2 \sqrt{2 \pi { \ensuremath{\widetilde{\alpha}({\ensuremath{\epsilon}})}}(1-{\ensuremath{\widetilde{\alpha}({\ensuremath{\epsilon}})}})}}{8 { \ensuremath{k}}(1 - 2 { \ensuremath{\epsilon } } ) p_x({\ensuremath{\theta^ * } } ) - 4\sqrt{2 \pi { \ensuremath{\widetilde{\alpha}({\ensuremath{\epsilon}})}}(1-{\ensuremath{\widetilde{\alpha}({\ensuremath{\epsilon } } ) } } ) } } \right].\end{aligned}\ ] ] in both cases , the asymptotic mse is minimal for @xmath112 . _
proof : @xmath119_if sensor node @xmath40 transmits a bit @xmath120 at round @xmath121 , then the fusion center receives the random variable @xmath122 where @xmath123 is bernoulli with parameter @xmath124 , and @xmath125 denotes addition modulo two .
since @xmath123 is independent of the transmitted bit ( which is bernoulli with parameter @xmath126 ) , the received value @xmath127 is also bernoulli , with parameter @xmath128 consequently , if we set @xmath111 according to equation , both algorithms will have their unique fixed point when @xmath129 , so will compute the @xmath60-quantile of @xmath14 .
the claimed form of the asymptotic variances follows from by performing calculations analogous to the proofs of theorems [ thmmbf ] and [ thm1bf ] .
in particular , the partial derivative with respect to @xmath77 now has a multiplicative factor @xmath130 , arising from equation and the chain rule . to establish that the asymptotic variance is minimized at @xmath112
, it suffices to note that the derivative of the mse with respect to @xmath124 is positive , so that it is an increasing function of @xmath124 .
@xmath131 of course , both the algorithms will fail , as would be expected , if @xmath132 corresponding to pure noise
. however , as summarized in proposition [ propnoise ] , as long as @xmath133 , feedforward noise does not affect the asymptotic rate itself , but rather only the pre - factor in front of the @xmath117 rate .
figure [ figlevels](b ) shows how the asymptotic variances @xmath134 and @xmath135 behave as @xmath124 is increased towards @xmath136 .
in this paper , we have proposed and analyzed different approaches to the problem of decentralized quantile estimation under communication constraints .
our analysis focused on the fusion - centric architecture , in which a set of @xmath0 sensor nodes each collect an observation at each time step .
after @xmath9 rounds of this process , the centralized oracle would be able to estimate an arbitrary quantile with mean - squared error of the order @xmath137 . in the decentralized formulation considered here ,
each sensor node is allowed to transmit only a single bit of information to the fusion center .
we then considered a range of decentralized algorithms , indexed by the number of feedback bits that the fusion center is allowed to transmit back to the sensor nodes . in the simplest case
, we showed that an @xmath138-bit feedback algorithm achieves the same asymptotic variance @xmath139 as the centralized estimator .
more interestingly , we also showed that that a @xmath12-bit feedback scheme , with suitably designed step sizes , can also achieve the same asymptotic variance as the centralized oracle .
we also showed that using intermediate amounts of feedback ( between @xmath12 and @xmath55 bits ) does not alter the scaling behavior , but improves the constant .
finally , we showed how our algorithm can be adapted to the case of noise in the feedforward links from sensor nodes to fusion center , and the resulting effect on the asymptotic variance .
our analysis in the current paper has focused only on the fusion center architecture illustrated in figure [ fignetwork ] .
a natural generalization is to consider a more general communication network , specified by an undirected graph on the sensor nodes .
one possible formulation is to allow only pairs of sensor nodes connected by an edge in this communication graph to exchange a bit of information at each round . in this framework
, the problem considered in this paper effectively corresponds to the complete graph , in which every node communicates with every other node at each round .
this more general formulation raises interesting questions as to the effect of graph topology on the achievable rates and asymptotic variances .
we would like to thank prof .
pravin varaiya for some discussion that led to the initial ideas on this subject .
rr was supported by the california department of transportation through the california path program .
mjw was partially supported by nsf grant dms-0605165 and an nsf career award ccf-0545862 . _
proof : @xmath119_first notice that by definition : @xmath142 \big ] , \ ] ] where @xmath14 is a @xmath143 random variable .
note that if @xmath144 , with @xmath145 , then certainly @xmath146meaning that @xmath147 stochastically dominates @xmath14 . for any constant @xmath148 , @xmath149
furthermore , by the quantizer is , by definition , a monotonically non - decreasing function .
consequently , a standard result on stochastic domination @xcite implies that @xmath150 .
differentiability follows from the definition of the function .
the finiteness of the variance of the quantization step is clear by construction ; more specifically , a crude upper bound is @xmath151 .
thus , analogous to the previous theorems , lemma [ fq_mon_dec ] is used to establish almost sure convergence .
now , some straightforward algebra using the results of lemma [ f_mon_rate ] shows that the partial derivative @xmath152 is @xmath153 - { \mathbb{e}}[x ] \ ; { \ensuremath{\mathbb{p}}}\left[x - s_{k+1 } \leq \frac{x}{{{\ensuremath{m } } } } \leq x - s_{k } \right ] \right \},\ ] ] this will be used next . to compute the asymptotic variance , we again exploit asymptotic normality ( see equation ) as before : @xmath154 & = & { \mathbb{e}}\left[x { \mathbb{i}}\left(-\sqrt{m}s_{k+1}\leq \frac{x-{\ensuremath{\alpha^*}}{{\ensuremath{m } } } } { \sqrt{m } } \leq -\sqrt{{{\ensuremath{m}}}}s_{k}\right)\right]\nonumber\\ & = & \sqrt{m}{\mathbb{e}}\left[(z+{\ensuremath{\alpha^*}}\sqrt{{{\ensuremath{m}}}}){\mathbb{i}}\left(-\sqrt{m}s_{k+1}\leq z \leq -\sqrt{{{\ensuremath{m}}}}s_{k}\right)\right]\nonumber\\ & = & \sqrt{m}{\mathbb{e}}\left[z { \mathbb{i}}\left(-\sqrt{m}s_{k+1}\leq z \leq -\sqrt{{{\ensuremath{m}}}}s_{k}\right)\right]+s\nonumber\\ & \rightarrow & -\sqrt{{{\ensuremath{m}}}}\int_{\sqrt{m}s_{k}}^{\sqrt{m}s_{k+1 } } z \frac{\exp\left(\frac{-z^2}{2a}\right)}{\sqrt{2\pi a } } dz+s \nonumber\\ s & { \ensuremath { : \ , = } } & { \mathbb{e}}[x]p({{\ensuremath{m}}}(x - s_{k+1})\leq x\leq { { \ensuremath{m}}}(x - s_{k}))\end{aligned}\ ] ] a side note is that if one chooses @xmath158 , we are guaranteed that at least one @xmath159 does not go to zero in a fixed quantizer ( i.e. a quantizer where the levels @xmath160 do not depend on @xmath55 ) .
but the correction factor expression , and as a matter of fact , the optimum quantization of gaussian , suggests that the levels @xmath161 scale as @xmath162 . in this case , the factor is a constant , independent of @xmath55 .
the rate with respect to @xmath55 is the same , independent of quantization .
it is clear from previous analysis that if the best quantizers are chosen @xmath171 .
obviously @xmath169 over the class of optimal quantizers is a decreasing function of @xmath172 .
|
we consider the following problem of decentralized statistical inference : given i.i.d .
samples from an unknown distribution , estimate an arbitrary quantile subject to limits on the number of bits exchanged . we analyze a standard fusion - based architecture , in which each of @xmath0 sensors transmits a single bit to the fusion center , which in turn is permitted to send some number @xmath1 bits of feedback . supposing that each of @xmath0 sensors receives @xmath2 observations , the optimal centralized protocol yields mean - squared error decaying as @xmath3)$ ] .
we develop and analyze the performance of various decentralized protocols in comparison to this centralized gold - standard .
first , we describe a decentralized protocol based on @xmath4 bits of feedback that is strongly consistent , and achieves the same asymptotic mse as the centralized optimum .
second , we describe and analyze a decentralized protocol based on only a single bit ( @xmath5 ) of feedback . for step sizes independent of @xmath6 , it achieves an asymptotic mse of order @xmath7 $ ] , whereas for step sizes decaying as @xmath8 , it achieves the same @xmath3)$ ] decay in mse as the centralized optimum .
our theoretical results are complemented by simulations , illustrating the tradeoffs between these different protocols .
* keywords : * decentralized inference ; communication constraints ; distributed estimation ; non - parametric estimation ; quantiles ; sensor networks ; stochastic approximation .
|
this work was supported by jisc under contract no .
we have introduced the idea of semantometrics in @xcite as a new class of metrics for evaluating research . as opposed to existing bibliometrics , webometrics , altmetrics , etc .
, semantometrics are not based on measuring the number of interactions in the scholarly communication network , but build on the premise that full - text is needed to assess the value of a publication . in @xcite
we have attempted to create the first semantometric measure based on the idea of measuring the progress of scholarly discussion .
our hypothesis states that the added value of publication _
p _ can be estimated based on the semantic distance from the publications cited by _
p _ to the publications citing _
p_. this hypothesis is based on the process of how research builds on the existing knowledge in order to create new knowledge on which others can build .
a publication , which in this way creates a bridge between what we already know and something new , which will people develop based on this knowledge , brings a contribution to science @xcite . until recently , it was still technically challenging for us to obtain an evaluation dataset on which properties of the contribution metric could be analysed . in this respect , we are now able to report on the first large - scale analysis of this metric .
the goal of our study was to understand the properties and behaviour of the semantometric contribution measure in comparison with established research evaluation metrics .
we chose to use citation counts obtained from the microsoft academic graph ( mag ) @xcite , as the representative of bibliometrics , usage data ( readership ) obtained from mendeley , as the representative of altmetrics and research articles aggregated by the open access connecting repositories ( core ) system as a representative sample for studying the characteristics of the contribution measure .
our experiments have been conducted on a dataset obtained by merging data from core , mag and mendeley . to assemble this dataset , we mapped dois of papers from core with mag . using mag
we then identified dois of papers citing and cited by the core papers .
finally , we used the dois to retrieve metadata , readership counts and primarily the titles and the abstracts using the mendeley api . by merging these three datasets
, we obtained a final dataset containing metadata , citation counts and reader counts of about 1.6 million open access papers .
additionally , we obtained metadata , including titles and abstracts of over 10 million papers which cite or are cited by the 1.6 million papers from core and are needed to calculate the contribution metric .
the main area of interest to us was the relation between the contribution measure and citation counts .
the reason for this was the prevalence of use of citation counts in research evaluation .
while using metrics based purely on citation counts has been subject to much criticism , these metrics still remain best known and most widely adopted .
the aim was not to find a perfect correlation with citation counts , but rather demonstrate how does the contribution measure behave in relation to the well - known metric .
we have first investigated the distributions of the three metrics , these are shown in figures [ fig : results ] [ c ] , [ fig : results ] [ f ] and [ fig : results ] [ i ] . as expected , the citation distribution ( figure [ fig : results ] [ f ] ) is a long tail ( power law ) distribution .
this is consistent with existing studies @xcite .
the readership distribution ( figure [ fig : results ] [ c ] ) exhibits the same properties as the citation distribution .
in contrast to the first two metrics , the contribution distribution ( figure [ fig : results ] [ i ] ) resembles a normal distribution . to confirm our data are consistent with previous studies
, we have investigated the relation between the citation and reader counts .
we found that the two metrics are slightly correlated with pearson @xmath0 .
a similarly strong correlation has been reported also by @xcite .
this correlation can also be seen when comparing the averaged values in figures [ fig : results ] [ a ] and [ fig : results ] [ d ] .
in contrast to the reader counts , we found no correlation between the citation counts and contribution ( pearson @xmath1 ) .
however , according to figures [ fig : results ] [ g ] and [ fig : results ] [ e ] we can see that when comparing averaged values the behaviour of the contribution metric is not random , instead it is clearly correlated with citation counts .
we can observe that publications with a citation score above a certain threshold achieve on average consistently higher contribution ( figure [ fig : results ] [ g ] ) .
although the standard deviation shows it is not always the case , the results suggest that publications with more than 25 citations are more likely to have higher contribution .
however , once a paper receives around 90 citations , higher citation counts do not lead on average to a higher contribution .
we think this is an interesting observation that is consistent with our perception of research quality .
one possible and highly simplified explanation could be that receiving around 90 citations is typically an indication of quality work .
higher citation counts then typically reflect the size of the target audience community ( impact ) rather than higher quality of the underlying research work .
this leads us to the conclusion that the contribution metric seems to capture different aspects of research performance than citation counts .
similarly as in the previous case , there is no correlation between the contribution measure and reader counts , which is confirmed by pearson @xmath2 . interestingly , while we observed a correlation between the averaged contribution and citation counts , there seems to be no such relation between averaged contribution and reader counts ( figures [ fig : results ] [ b ] and [ fig : results ] [ h ] ) .
we have demonstrated that new measures for assessing publication impact , which take into account the manuscript of the publication , can be developed and presented a comparative study of the semantometric contribution measure with citation and reader counts .
the results of our study suggest that the contribution metric captures different aspects of research performance than citation counts .
more specifically , we believe that semantometrics have the potential to capture research quality and contribution rather than research impact .
|
over the recent years , there has been a growing interest in developing new research evaluation methods that could go beyond the traditional citation - based metrics .
this interest is motivated on one side by the wider availability or even emergence of new information evidencing research performance , such as article downloads , views and twitter mentions , and on the other side by the continued frustrations and problems surrounding the application of purely citation - based metrics to evaluate research performance in practice .
semantometrics are a new class of research evaluation metrics which build on the premise that full - text is needed to assess the value of a publication .
this paper reports on the analysis carried out with the aim to investigate the properties of the semantometric _ contribution _ measure @xcite , which uses semantic similarity of publications to estimate research contribution , and provides a comparative study of the contribution measure with traditional bibliometric measures based on citation counting .
< ccs2012 > < concept >
< concept_id>10010147.10010178.10010179</concept_id > < concept_desc > computing methodologies natural language processing</concept_desc > < concept_significance>500</concept_significance >
< /concept > < concept > < concept_id>10002951.10003317</concept_id > < concept_desc > information systems information retrieval</concept_desc > < concept_significance>300</concept_significance > < /concept >
< /ccs2012 >
|
a classical result in representation theory of quivers is gabriel s theorem , stating that a connected quiver supports only finitely many irreducible representations ( i.e. , of indecomposable modules over the associated path algebra ) if and only if it is of type @xmath0 , @xmath1 , @xmath2 .
the classification of tame quivers as _ euclidean graphs _ , or _ extended dynkin diagrams _ , of type @xmath3 , @xmath4 , @xmath5 came shortly afterwards . remarkably , any other quiver supports arbitrarily large families of indecomposable representations , i.e. , to be of _ wild representation type_. in algebraic geometry , the relevant problem in representation theory of algebras concerns the complexity of the category of maximal cohen - macaulay modules over the coordinate ring @xmath6 $ ] of a closed subvariety @xmath7 defined over an algebraically closed field @xmath8 . for @xmath9 , assuming @xmath6 $ ] to be cohen - macaulay ( so @xmath10 is said to be arithmetically cohen - macaulay , briefly acm ) , these in turn correspond to acm sheaves , namely coherent sheaves @xmath11 on @xmath10 without intermediate cohomology i.e. satisfying @xmath12 for all @xmath13 and @xmath14 . for hypersurfaces , in view of @xcite these modules also correspond to matrix factorizations , which in turn are related to mirror symmetry , cf .
@xcite . in this sense , reduced projective acm varieties of finite cm type have been classified in @xcite , cf . also @xcite . their list ( for positive dimension )
consists of rational normal curves , projective spaces , smooth quadrics , the veronese surface in @xmath15 and the cubic scroll in @xmath16 .
this ties in with grothendieck s classical result on the splitting of vector bundles over @xmath17 , @xcite , in turn rooted in ideas of segre , @xcite . our goal here is to tell what happens to other varieties .
smooth cubic surfaces are of wild cm type according to the main point of @xcite .
some more varieties have been proved to fall in this class , for instance all segre varieties besides the cm finite ones ( cf .
@xcite ) , del pezzo surfaces ( cf .
@xcite ) , positive - dimensional hypersurfaces of degree at least @xmath18 and some complete intersections ( cf .
@xcite ) some fano varieties ( cf .
@xcite ) , the triple veronese embedding of any variety ( cf .
@xcite ) . as for cm tame projective varieties ,
besides smooth elliptic curves ( by seminal work of atiyah , @xcite , also related to classical work of segre , cf @xcite ) and rational chains of type @xmath3 ( cf .
@xcite , cf .
also @xcite ) , the only known examples were given in @xcite and consist of smooth quartic surface scrolls in @xmath15 .
here we prove the following result .
let @xmath19 be a closed integral acm subvariety of positive dimension which is not a cone .
then @xmath10 is of wild cm type unless @xmath10 is one of the following : a. [ linear ] a linear subspace ; b. [ quadric ] a smooth quadric ; c. [ rnc ] a rational normal curve ; d. [ elliptic ] a smooth or nodal curve of arithmetic genus @xmath20 ; e. [ scroll ] a smooth rational surface scroll of degree @xmath21 ; f. [ veronese ] the veronese surface in @xmath15 .
our result implies a strong version of the finite - tame - wild trichotomy for the class of varieties appearing in the previous theorem , namely that any such variety @xmath10 falls in exactly one of the following classes : * there are only finitely many indecomposable acm sheaves over @xmath10 up to isomorphism .
this happens in cases , , , for @xmath22 and .
* for any given rank @xmath23 , the parameter space of indecomposable non - isomorphic acm sheaves of rank at most @xmath23 consists of the union of finitely many isolated points and connected curves of genus @xmath24 ( in case and @xmath25 ) or @xmath20 ( in case ) .
* @xmath10 is of wild cm type in the algebraic sense and hence in the geometric sense , i.e. @xmath10 admits families of arbitrarily large dimension of indecomposable non - isomorphic acm sheaves . it should be noted that , although our assumptions on the variety @xmath10 are rather mild , a different behaviour might be expected when relaxing them .
for instance a quadric cone of corank @xmath20 is cm countable , cf .
@xcite ; several other examples of this kind exist over local rings , cf .
for instance @xcite for a detailed picture , see also @xcite .
more varieties of tame type appear from germs of elliptic singularities , cf .
@xcite or from non - isolated affine surface singularities , see also @xcite .
let us indicate the strategy of our proof .
let @xmath26 be a variety of the class under examination : an integral closed acm subscheme of @xmath27 which is not a cone .
the first step is to isolate some essential datum in order to build large families of indecomposable non - isomorphic mcm modules on @xmath28 $ ] . in theorem
[ strictly wild by extensions ] we establish that this should be a pair of acm sheaves @xmath29 and @xmath30 on @xmath26 , whose only endomorphisms are homotheties ( i.e. @xmath29 and @xmath30 are _ simple _ ) , having no non - trivial maps in either direction @xmath31 or @xmath32 and such that the extension space @xmath33 is sufficiently large , namely of dimension @xmath34 .
indeed , from this datum we construct a fully faithful exact functor from the category @xmath35 of finite - dimensional representations of the kronecker quiver @xmath36 with @xmath37 arrows to the category @xmath38}$ ] of mcm modules over @xmath28 $ ] .
this quiver should be seen as parametrizing acm sheaves appearing as extension of copies of @xmath29 and @xmath30 . in turn , by a standard argument , the existence of such functor suffices to prove cm wildness of @xmath26 .
however , constructing the sheaves @xmath29 and @xmath30 turns out to be quite complicated in general .
for this we need our next result , theorem [ stable ff ] , which shows how to deduce cm wildness of @xmath10 from cm wildness of @xmath26 when @xmath26 is a linear section of @xmath10 of codimension @xmath39 . for this
, we need to further assume that @xmath29 and @xmath30 are ulrich sheaves , i.e. their modules of global sections have the maximal number of generators .
we see this as a further prove of the importance of these sheaves , cf .
we also need the extra assumption @xmath40 , which turns out to hold everywhere except for varieties of minimal degree , where in turn we do not need it .
the idea of theorem [ stable ff ] is that taking the @xmath41-th syzygy @xmath42}$ ] of the @xmath6$]-module of an mcm module @xmath43 over @xmath28 $ ] one obtains an mcm module over @xmath6 $ ] and that this entails no essential loss of information if @xmath43 is ulrich .
in fact , for @xmath44}^c$ ] to be a functor we need to pass to the stable category @xmath45}$ ] where we quotient out by morphisms factoring through a free module .
the point is that the stable syzygy functor @xmath46 is fully faithful on ulrich modules .
the proof of this fact relies on a duality argument and on cohomology vanishing of ulrich sheaves .
the next result , theorem [ embedding ] , shows how to put these two ingredients together .
indeed , by resolving over @xmath6 $ ] the module of global sections of the universal extension of the sheaves @xmath30 by @xmath29 over @xmath26 needed for theorem [ strictly wild by extensions ] and taking its @xmath41-th syzygy , we get a functor @xmath47 whose stabilization is fully faithful by theorem [ stable ff ] .
then the functor itself is not quite fully faithful , nevertheless is a _ representation embedding _ , i.e. it sends non - isomorphic ( resp .
irreducible ) representations to non - isomorphic ( resp .
indecomposable ) modules , and this suffices to show cm wildness of @xmath10 . summing up , it only remains to construct the ulrich sheaves @xmath29 and @xmath30 as above over a linear section @xmath26 of @xmath10 , which we take to be of dimension @xmath20 when the sectional genus @xmath48 of @xmath10 is at least @xmath49 , or of dimension @xmath49 for @xmath50 .
this is the content of [ ony ] , which we now review .
we first take care of the case @xmath51 , i.e. of varieties of minimal degree . here
, our assumptions combined with del pezzo and bertini s classification say that @xmath10 is a rational normal scroll or the veronese surface in @xmath15 , and these varieties are easily dealt with .
note that here is where all cm - finite cases appear , together with the two sporadic cases given by the two rational quartic surface scrolls .
next , the case @xmath52 is quite easily seen to provide only cm wild varieties , as @xmath29 and @xmath30 can be taken to be sufficiently general bundles of rank @xmath49 over @xmath26 of slope @xmath53 . finally , in the case @xmath54 our proof of the existence of @xmath29 and @xmath30 is based on a detailed study of ulrich sheaves on surfaces of almost minimal degree .
when @xmath26 smooth ( i.e. @xmath26 is a del pezzo surface ) or even just normal , this can be carried out via the well - known hartshorne - serre construction associated with general sets of @xmath55 points of @xmath26 . on the other hand , when @xmath26 is not normal , we take @xmath29 and @xmath30 to be ulrich sheaves of rank @xmath20 which are not locally free , in fact the main point is to bound from below the number of independent global sections of the first local @xmath56 sheaf between @xmath29 and @xmath30 .
to accomplish this , according to the classification of these surfaces , it suffices to settle a finite number of cases . in the main one ( when @xmath26 is normalized by a scroll ) we define @xmath29 and @xmath30 to be direct images of certain line bundles via the normalization map @xmath57 , and reduce the result to counting the intersection points of the moving part of the associated curves .
let @xmath8 be an algebraically closed field , and let @xmath58 $ ] be the graded polynomial ring in @xmath59 indeterminates .
write @xmath6 $ ] for the homogeneous coordinate ring of a closed subscheme @xmath10 of @xmath60 , and @xmath61 for the restriction to @xmath10 of @xmath62 .
we first recall some basic terminology for various cohen - macaulay properties of varieties , sheaves and modules .
let @xmath7 be a closed subscheme of dimension @xmath63 .
then @xmath10 is arithmetically cohen - macaulay ( acm ) if the ring @xmath6 $ ] of @xmath10 is cohen - macaulay , i.e. if @xmath6 $ ] has a graded @xmath64-free resolution of length @xmath65 . given a closed subscheme @xmath7 , we set @xmath66 $ ] and let @xmath67 be the category of finitely generated maximal cohen - macaulay ( mcm ) modules over @xmath68 . if @xmath10 is an acm variety of positive dimension , then the sheafification functor provides an equivalence between @xmath67 and the full subcategory of @xmath69 consisting of arithmetically cohen - macaulay ( acm ) sheaves , i.e. coherent sheaves whose @xmath68-module of global sections is mcm . given @xmath70 in @xmath67 of positive rank ,
the number of independent minimal generators of @xmath70 is at most @xmath71 . for the corresponding sheaf @xmath11 over @xmath10 ,
assuming @xmath72 , we have : @xmath73 we say that @xmath11 is an _
ulrich sheaf _ on @xmath10 and that @xmath70 is an ulrich module over @xmath68 if equality is attained in .
put @xmath74 for the full subcategory of @xmath67 consisting of ulrich modules .
we say that a coherent sheaf @xmath11 over @xmath10 is _ simple _ if its only endomorphisms are homotheties . given two coherent sheaves @xmath11 and @xmath75 over @xmath10 , we write @xmath76 for the alternating sum of the dimension of @xmath77 .
we abbreviate @xmath78 .
we will consider a couple of related notions of cm wildness of a variety .
algebraically this means that , for any finitely generated associative @xmath8-algebra @xmath79 , the category of mcm modules over @xmath66 $ ] contains , in some sense , the category @xmath80 of left @xmath79-modules which have finite dimension over @xmath8 .
let @xmath7 be an equidimensional closed subscheme and let @xmath68 .
for any finitely generated associative @xmath8-algebra @xmath79 , and any @xmath81-bimodule @xmath82 , flat over @xmath79 , define the functor : @xmath83 the variety @xmath10 is of _ wild cm type _ if , for any @xmath79 as above , there is @xmath82 such that @xmath84 takes values in @xmath67 and is a _ representation embedding _ ,
i.e. : a. the module @xmath85 is decomposable whenever @xmath86 is ; b. for any pair @xmath87 of modules in @xmath80 , we have : @xmath88 the variety @xmath10 is of _ wild ulrich type _ if moreover : [ [ a ) ] ] c. for any @xmath85 in @xmath80 , @xmath86 is ulrich . the variety @xmath10 is said to be _ strictly cm wild
_ if for any @xmath79 as above there is @xmath82 such that @xmath84 takes values in @xmath67 and is fully faithful , i.e. : @xmath89 if moreover @xmath86 is ulrich for all @xmath85 , then @xmath10 is _ strictly ulrich wild_. [ well known ] the following facts are well - known .
a. if the subscheme @xmath10 is strictly cm wild , then it is of wild cm type , and if @xmath10 is strictly ulrich wild then it is of wild ulrich type .
b. if @xmath10 is of wild cm type , then for any @xmath90 there are families of dimension at least @xmath23 consisting of indecomposable acm sheaves on @xmath10 , all non - isomorphic to one another . in other words
, @xmath10 is of wild cm type in the geometric sense .
if @xmath10 is of wild ulrich type , these families can be taken to consist of ulrich sheaves . c. any fully faithful exact functor @xmath91 is of the form @xmath84 for some @xmath79-flat @xmath81-bimodule @xmath82 .
let @xmath92 be an integer and consider the kronecker quiver @xmath93 with two vertices and @xmath37 arrows from the first vertex to the second .
write @xmath35 for the abelian category of finite - dimensional @xmath8-representations of @xmath94 .
[ [ i ) ] ] d. to check that @xmath10 is strictly cm wild ( resp . , of wild cm type ) , it suffices to construct a fully faithful exact functor ( resp . , a representation embedding ) : @xmath95 where @xmath93 is the kronecker quiver with @xmath34 .
if moreover @xmath96 is ulrich for any @xmath97 in @xmath35 , then @xmath10 is strictly ulrich wild ( resp .
, of wild ulrich type ) .
one of the main tools that we use to build large families of mcm modules ( or acm sheaves ) is to start with two such sheaves and build extensions in a systematic way .
this goes as follows .
let @xmath19 be a closed subscheme , and let @xmath29 and @xmath30 be two coherent sheaves on @xmath10 such that : @xmath98 set @xmath99 and consider the projective space @xmath100 of lines through the origin in @xmath101 .
then , over @xmath102 , there is a universal extension : @xmath103 where we write @xmath48 and @xmath104 for the projections from @xmath102 to @xmath10 and @xmath100 , and for @xmath105 and @xmath106 , we set @xmath107 . then we consider : @xmath108 it is clear that : @xmath109.\ ] ] set @xmath110 and consider the kronecker quiver @xmath111 .
then , the natural isomorphism @xmath112 provides an equivalence : @xmath113 explicitly , this is described as follows .
choose a basis @xmath114 of @xmath99 .
let @xmath97 be a representation of @xmath94 having dimension vector @xmath115 .
then @xmath97 corresponds to the choice of @xmath37 linear maps @xmath116 .
take the element : @xmath117 then , under @xmath112 , we obtain from @xmath118 a morphism : @xmath119 the cone of @xmath70 is the element of @xmath120 associated with @xmath97 via @xmath121 .
this is directly extended to morphisms .
[ strictly wild by extensions ] let @xmath7 be a closed subscheme and let @xmath29 and @xmath30 be simple coherent sheaves on @xmath10 satisfying : @xmath122 set @xmath110 , assume @xmath123 , and put @xmath111 .
then the restriction @xmath124 of @xmath125 to @xmath35 is a fully faithful exact functor : @xmath126 let us first give a more explicit description of @xmath124 . at the level of objects , given a representation @xmath97 of the quiver @xmath35 with dimension vector @xmath115
let @xmath127 be the @xmath37 linear maps associated with @xmath97 and let @xmath118 be as in .
then @xmath96 fits as middle term of a representative of the extension class corresponding to @xmath118 : @xmath128 let us now check that this is well - defined on morphisms .
let @xmath129 be another representation of @xmath94 , of dimension vector @xmath130 , corresponding to the linear maps @xmath131 .
a morphism @xmath132 of representations is given by linear maps @xmath133 and @xmath134 such that : @xmath135 recall that @xmath136 , and consider the map of coherent sheaves @xmath137 .
then , it is well - known that this map defines a morphism of extensions : @xmath138 & { \mathscr{a}}^a \ar[d]^-{\beta_{\mathscr{a } } } \ar[r ] & \phi({\mathscr{r } } ) \ar[d]^{\phi } \ar[r ] & { \mathscr{b}}^b \ar@{=}[d ] \ar[r ] & 0\\ 0 \ar[r ] & { \mathscr{a}}^c \ar^-i[r ] & { \mathscr{d}}\ar^p[r ] & { \mathscr{b}}^b \ar[r ] & 0 } \ ] ] for a certain sheaf @xmath139 representing the element : @xmath140 analogously @xmath141 defines : @xmath142 & { \mathscr{a}}^c \ar@{=}[d ] \ar^{i'}[r ] & { \mathscr{d } } ' \ar^-{\phi'}[d ] \ar^{p'}[r ] & { \mathscr{b}}^b \ar[d]^{\alpha_{\mathscr{b } } } \ar[r ] & 0\\ 0 \ar[r ] & { \mathscr{a}}^c \ar[r ] & \phi({\mathscr{s } } ) \ar[r ] & { \mathscr{b}}^d \ar[r ] & 0 } \ ] ] with upper row representing an extension class in : @xmath143 now we can use to see that lower extension of is the same as the upper one .
then the morphisms @xmath144 and @xmath145 of extensions compose to map from @xmath96 to @xmath146 .
it is clear that this construction agrees with the definition of @xmath124 .
indeed , the representation @xmath97 is mapped by @xmath121 to the cone of the matrix @xmath70 of , which in turn is sent by @xmath147 to the cone of : @xmath148 \to { \mathscr{a}}^a.\ ] ] by construction this cone is represented by the extension class @xmath96 .
now it only remains to use this setup to show that the functor @xmath124 is fully faithful . to check this ,
consider a commutative diagram : @xmath149 & { \mathscr{a}}\ar[d]^-{0 } \ar^{i}[r ] & { \mathscr{d}}\ar[d]^-{\lambda } \ar[r]^{p } & { \mathscr{b}}\ar[d]^-{0 } \ar[r ] & 0\\ 0 \ar[r ] & { \mathscr{a}}\ar[r]^{i ' } & { \mathscr{d } } ' \ar[r]^{p ' } & { \mathscr{b}}\ar[r ] & 0 } \ ] ] since @xmath150 , we have @xmath151 . but
@xmath152 implies that @xmath153 factors as : @xmath154^{\lambda } \ar[dr ] & & & { \mathscr{a}}\\ & { \mathscr{b}}\ar@{=}[r ] & { \mathscr{d}}/{\mathscr{a}}\ar[ur]_-{\bar{\lambda } } & } \ ] ] if @xmath155 , this would give a nonzero map @xmath156 , contradicting @xmath157 . with this in mind , it is easy to deduce injectiveness of the natural map : @xmath158 as for surjectiveness , given a morphism @xmath159 , we compose @xmath160 on one side with the projection @xmath161 , and with the injection @xmath162 on the other side .
we obtain thus a map @xmath163 , which must vanish since @xmath164 .
we deduce that @xmath160 defines maps @xmath165 and @xmath166 , which must be of the form @xmath167 and @xmath168 by the assumption that @xmath29 and @xmath30 are simple .
it is now clear that the pair @xmath169 defines a morphism @xmath170 whose image under is @xmath160 .
we deduce a criterion for an acm variety being strictly cm wild . in the hypothesis of theorem
[ strictly wild by extensions ] , we have : a. if @xmath34 and @xmath29 and @xmath30 are acm , then @xmath10 is strictly cm wild ; b. if moreover @xmath29 and @xmath30 are ulrich , then @xmath10 is strictly ulrich wild . by construction of the functor @xmath124 of theorem
[ strictly wild by extensions ] , the sheaf @xmath96 associated with a representation @xmath97 of @xmath94 is acm ( respectively , ulrich ) if @xmath29 and @xmath30 are acm ( respectively , ulrich ) .
we conclude using remark [ well known ] .
the hypothesis @xmath157 in theorem [ strictly wild by extensions ] is necessary . indeed , if @xmath171 we could consider the map : @xmath172 this map @xmath145 is not zero , and makes the following diagram commute : @xmath173 & { \mathscr{a}}\ar[d]^-{0 } \ar^{i}[r ] & { \mathscr{d}}\ar[d]^-{\phi } \ar[r]^{p } & { \mathscr{b}}\ar[d]^-{0 } \ar[r ] & 0\\ 0 \ar[r ] & { \mathscr{a}}\ar[r]^{i ' } & { \mathscr{d}}\ar[r]^{p ' } & { \mathscr{b}}\ar[r ] & 0 } \ ] ] notice , however that any @xmath145 fitting in such a diagram will be nilpotent . as an explicit example , let @xmath174 be an hypersurface of degree @xmath175 and @xmath176 an ag subscheme of codimension two and index @xmath177 , where : @xmath178 define @xmath179 and assume that @xmath180 so that @xmath181 .
let @xmath139 be the non - trivial extension of @xmath182 by @xmath183 ( one can show that this extension exists and is unique up to a nonzero scalar , by the definition of @xmath184 and by serre duality ) .
when @xmath185 is not a complete intersection inside @xmath10 , @xmath139 is indecomposable .
anyway @xmath139 is an acm sheaf of rank @xmath49 over @xmath10 which is never simple , as it always admits a nilpotent endomorphism .
the conclusion of theorem [ strictly wild by extensions ] clearly fails in this case .
let @xmath7 be a closed acm subscheme of dimension @xmath186 , let @xmath26 be a linear section of @xmath10 of codimension @xmath187 .
set @xmath188 $ ] , @xmath66 $ ] and write @xmath189 for the dualizing sheaf of @xmath26 .
looking at a finitely generated graded module @xmath43 over @xmath190 as a graded module over @xmath68 , we take its minimal graded @xmath68-free resolution : @xmath191 write @xmath192 for the @xmath193-th syzygy of @xmath43 over @xmath68 , i.e. @xmath194 .
it is well - known that , if @xmath43 is mcm over @xmath190 , then @xmath195 is mcm over @xmath68 for @xmath196 .
let @xmath197 the stable category of maximal cohen - macaulay modules over @xmath68 .
given @xmath198 in @xmath67 , we write @xmath199 for the morphisms in the this category , namely the morphisms of degree @xmath24 from @xmath70 to @xmath200 , modulo the ideal of morphisms ( of degree zero ) that factor through a free @xmath68-module .
we write @xmath201 for the stabilization functor : @xmath202 for @xmath196 , we have also the @xmath193-th syzygy stable functor : @xmath203 [ stable ff ] assume @xmath204 . then the restriction to @xmath205 of the @xmath41-th stable syzygy functor provides a fully faithful embedding : @xmath206
the following lemma will be one of the keys of our analysis . given a finitely generated graded @xmath68-module @xmath70 , we write @xmath207 for the graded submodule of @xmath70 generated by the elements of degree at most @xmath175 of @xmath70 .
we also write @xmath208 .
[ degree ] fix the hypothesis as in theorem [ stable ff ] , let @xmath43 be an ulrich module over @xmath190 , and set @xmath209 .
then we have a functorial exact sequence : @xmath210 recall that @xmath43 is generated in a single degree , and that the number of minimal generators of @xmath43 equals @xmath211 .
in other words , we can assume @xmath212 . by the minimality of the resolution , we have , for @xmath213 : @xmath214 also , @xmath10 and @xmath26 are not linear subspaces , so that @xmath43 has no free summands .
the coherent sheaf over the variety @xmath10 associated with @xmath43 is of the form @xmath215 , where @xmath216 is an ulrich sheaf over @xmath26 and @xmath217 is the obvious inclusion . then sheafifying we get an exact complex : @xmath218 where @xmath219 .
set @xmath220 for the sheafification of @xmath70 .
now apply @xmath221 to and use @xmath222 for @xmath223 .
note also that grothendieck duality gives : @xmath224 we obtain a long exact sequence : @xmath225 next , we apply @xmath226 to this sequence and use that @xmath10 is cohen - macaulay so that @xmath227 .
also , we easily see that : @xmath228 as the first isomorphism is obvious , while the second one follows from a standard exchange argument of homomorphism groups .
moreover , since @xmath26 is cut in @xmath10 by @xmath41 hyperplanes forming a regular sequence , adjunction theory says that @xmath229 .
therefore , using @xmath230 we get : @xmath231 putting together , we get the long exact sequence : @xmath232 taking global sections of this sequence , we just get the result of applying @xmath233 to , i.e. the long exact sequence : @xmath234 now comes the main point , namely that @xmath235 . to see this , recall that @xmath236 and that @xmath216 is ulrich on @xmath26 as well as @xmath237
then , by ( * ? ? ?
* proposition 2.1 ) we get : @xmath238 note that , by the assumption @xmath204 we get an embedding : @xmath239 therefore , @xmath240 implies @xmath241 .
we have thus established that @xmath242 contains no element of degree @xmath243 .
also , we may write : @xmath244 then , says that @xmath245 generates all the elements of @xmath246 of degree at most @xmath247 , i.e. the image of @xmath248 is the submodule @xmath249 of @xmath246 .
this is clearly functorial , and the lemma is now proved .
let @xmath43 and @xmath85 be two ulrich modules over @xmath190 .
our goal will be to describe two mutually inverse maps : @xmath250 set @xmath251 and @xmath252 .
first , let @xmath253 belong to @xmath254 .
consider the minimal graded free resolutions of @xmath43 and @xmath85 over @xmath68 and choose a lifting of @xmath255 to these resolutions : @xmath256 l \ar^\varphi [ d ] & \ar[l ] f_0 \ar^{\varphi_0}[d ] & \ar[l ] \cdots & \ar[l ] f_{c-1 } \ar^{\varphi_{c-1}}[d ] & \ar[l ] m \ar@{.>}^{\tilde \varphi}[d]\\ 0 & \ar[l ] n & \ar[l ] g_0 & \ar[l ] \cdots & \ar[l ] g_{c-1 } & \ar[l ] p } \ ] ] the morphism @xmath257 induced on the @xmath41-th syzygy modules gives the class @xmath258 in @xmath259 .
this does not depend on the choice of the lifting @xmath260 , as any other choice would provide a map @xmath261 such that @xmath262 factors through a free module .
conversely , given @xmath263 , we choose a representative @xmath264 and take its dual @xmath265 . since @xmath266 is homogeneous of degree @xmath24 , it maps elements of degree at most @xmath247 in @xmath267 to elements of degree at most @xmath247 in @xmath246 . by lemma [ degree ]
we obtain a diagram : @xmath268 & \langle m^*_{\le { 1-c } } \rangle \ar[r ] & m^ * \ar[r ] & \operatorname{hom}_t(l , t(c ) ) \ar[r ] & 0\\ 0 \ar[r ] & \langle p^*_{\le { 1-c } } \ar_-{\psi^*}[u ] \rangle \ar[r ] & p^ * \ar[r ] \ar_-{\psi^*}[u ] & \operatorname{hom}_t(n , t(c ) ) \ar_-{\hat \psi}@{.>}[u]\ar[r ] & 0 } \ ] ] we wish to associate with @xmath269 the morphism @xmath270 .
to do this , we have to check that @xmath271 does not depend on the choice of the representative @xmath272 of @xmath269 . by definition
any other representative differs from @xmath272 by a map @xmath273 that factors through a free module , which we call @xmath274 , i.e. @xmath275 with @xmath276 and @xmath277 .
therefore @xmath278 factors through @xmath279 and again by lemma [ degree ] we get the commutative diagram : @xmath173 & \langle m^*_{\le { 1-c } } \rangle \ar[r ] & m^ * \ar[r ] & \operatorname{hom}_t(l , t(c ) ) \ar[r ] & 0\\ 0 \ar[r ] & \langle f^*_{\le { 1-c } } \rangle \ar_-{\zeta_1^*}[u ] \ar[r ] & f^*\ar_-{\zeta_1^*}[u ] \\ 0 \ar[r ] & \langle p^*_{\le { 1-c } } \rangle \ar_-{\zeta_2^*}[u ] \ar[r ] & p^ * \ar[r ] \ar_-{\zeta_2^*}[u ] & \operatorname{hom}_t(n , t(c ) ) \ar^-{\hat \zeta}@{.>}[uu]\ar[r ] & 0 } \ ] ] call @xmath280 the quotient @xmath281 .
the diagram says that @xmath282 factors through @xmath280 . now observe that @xmath280 is a free module .
indeed , any direct summand of @xmath274 takes the form @xmath283 for some @xmath284 , and : @xmath285 therefore @xmath280 is the direct sum of all summands @xmath283 of @xmath279 with @xmath286 , hence @xmath280 is free . but
@xmath287 is a torsion module , so it admits no non - trivial morphism with target in @xmath280 , and therefore @xmath288 .
let us now check that these maps are mutually inverse .
given @xmath289 , we consider a representative @xmath290 of the class @xmath258 .
dualizing we obtain a commutative diagram : @xmath291 & f_{c-1}^ * \ar[r ] & m^ * \ar[r ] & \operatorname{hom}_t(l , t(c ) ) \ar[r ] & 0\\ \cdots \ar[r ] & g_{c-1}^ * \ar_-{\varphi_{c-1}^*}[u ] \ar[r ] & p^ * \ar[r ] \ar_-{\psi^*}[u ] & \operatorname{hom}_t(n , t(c ) ) \ar_-{\varphi^*}@{.>}[u]\ar[r ] & 0 } \ ] ] this diagram is the extension of to a minimal resolution of @xmath292 and @xmath293 .
this says that @xmath294 , so @xmath295 .
conversely , let @xmath272 be a representative of @xmath296 and set @xmath297 .
lift the map @xmath298 induced by @xmath272 to the minimal graded free resolutions of these modules and dualize to obtain : @xmath299 l \ar^\varphi [ d ] & \ar[l ] f_0 \ar^{\psi_0}[d ] & \ar[l ] \cdots & \ar[l ] f_{c-1 } \ar^{\psi_{c-1}}[d ] & \ar[l ] m \ar^{\psi}[d ] & \ar[l ] 0\\ 0 & \ar[l ] n & \ar[l ] g_0 & \ar[l ] \cdots & \ar[l ] g_{c-1 } & \ar[l ] p&\ar[l ] 0 } \ ] ] then @xmath272 is induced by a lifting of @xmath253 to the minimal resolutions of @xmath43 and @xmath85 .
our proof is thus complete .
we isolate the following easy consequence of lemma [ degree ] .
[ not free ] if @xmath43 is an ulrich module @xmath43 over @xmath190 , then @xmath300 has no free summands .
suppose @xmath301 , with @xmath274 a non - zero free direct summand .
then by lemma [ degree ] we have : @xmath302 with @xmath248 diagonal .
then , says that the restriction of @xmath248 is an isomorphism between @xmath303 and @xmath274 , as obviously @xmath43 has no free summand .
therefore , looking at we see that @xmath304 is surjective onto @xmath274 , which contradicts minimality of the resolution .
let @xmath305 be a hypersurface of degree @xmath175 and @xmath26 be a linear section of codimension @xmath41 of @xmath10 .
then , by matrix factorization the minimal graded free resolution of an ulrich module @xmath43 of rank @xmath23 on @xmath188 $ ] reads : @xmath306 since @xmath307 , this yields a resolution : @xmath308 combining with the koszul resolution of @xmath26 in @xmath10 we get a resolution over @xmath66 $ ] : @xmath309 the @xmath310-th term @xmath311 of this resolution looks as follows ( here @xmath312 ) : @xmath313 let @xmath251 .
the resolution of the dualized syzygy @xmath246 starts with : @xmath314 we may now remove from this the dual of the truncation at @xmath251 of , which is to say , by lemma [ degree ] , the resolution of @xmath315 .
the residual strand recovers precisely , twisted by @xmath316 .
the two strands of the resolution do not mix if @xmath317 .
the assumption @xmath204 is crucial , as shown by the following example .
take @xmath318 and let @xmath26 be a cubic scroll obtained as hyperplane section of @xmath10 .
note that in this case @xmath319 .
if @xmath274 is a line in @xmath26 and @xmath320 is the class of a hyperplane in @xmath26 , then @xmath321 is an ulrich line bundle on @xmath26 , whose sheafified minimal @xmath183-resolution reads : @xmath322 the sheafified first syzygy @xmath220 of @xmath216 has a resolution : @xmath323 in this case , @xmath324 appears as cokernel of a map @xmath325 , but this can not be singled out by degree reasons only from the first map of the above resolution .
let us fix the setup for this section .
let @xmath19 be an acm subscheme , put @xmath66 $ ] let and @xmath26 be a linear section of @xmath10 of codimension @xmath41 and dimension @xmath186 .
let @xmath29 and @xmath30 be two simple ulrich sheaves on @xmath26 .
set @xmath326 , @xmath327 , and assume @xmath123 .
put @xmath93 . over @xmath328 , there is a universal extension : @xmath329 take the sheafified minimal graded free resolutions of @xmath29 and @xmath30 as @xmath183-modules , pull - back via @xmath48 to @xmath102 , and use the mapping cone construction to build a minimal graded free resolution of @xmath330 over @xmath102 : @xmath331 & { \mathscr{a}}\boxtimes { \mathscr{o}}_{{\mathds p}w}\ar[u ] \ar[r ] & { \mathscr{u}}\ar[u]\ar[r ] & { \mathscr{b}}\boxtimes { \mathscr{o}}_{{\mathds p}w}(-1 ) \ar[r ] \ar[u ] & 0 \\ 0 \ar[r ] & { \mathscr{f}}_0 \ar[u ] \boxtimes { \mathscr{o}}_{{\mathds p}w } \ar[r ] & { \mathscr{h}}_0 \ar^-{d_0}[u ] \ar[r ] & { \mathscr{g}}_0 \ar[u ] \boxtimes
{ \mathscr{o}}_{{\mathds p}w}(-1 ) \ar[r ] & 0 \\ & \vdots \ar[u ] & \vdots \ar[u ] & \vdots \ar[u]\\ 0 \ar[r ] & { \mathscr{f}}_{c-1 } \ar[u ] \boxtimes { \mathscr{o}}_{{\mathds p}w } \ar[r ] & { \mathscr{h}}_{c-1 } \ar[u ] \ar[r ] & { \mathscr{g}}_{c-1 } \ar[u ] \boxtimes { \mathscr{o}}_{{\mathds p}w}(-1 ) \ar[r ] & 0 \\ 0 \ar[r ] & { \mathscr{f}}_{c } \ar[u ] \boxtimes { \mathscr{o}}_{{\mathds p}w } \ar[r ] & { \mathscr{h}}_{c } \ar^-{d_c}[u ] \ar[r ] & { \mathscr{g}}_{c } \ar[u ] \boxtimes { \mathscr{o}}_{{\mathds p}w}(-1 ) \ar[r ] & 0 \\ & \vdots \ar[u ] & \vdots \ar[u ] & \vdots \ar[u ] } \ ] ] here , @xmath332 .
set : @xmath333 then we consider : @xmath334 the following lemma is clear .
let @xmath188 $ ] , and set @xmath43 and @xmath85 for the modules of global sections of @xmath29 and @xmath30 .
let @xmath220 and @xmath335 be the sheafifications of @xmath336 and @xmath337 . then : @xmath338.\ ] ] now consider the equivalence @xmath121 of .
then the restriction of @xmath339 to @xmath35 , composed with the global sections functor , gives an exact functor : @xmath340 [ embedding ] assume @xmath341 and @xmath204 .
then @xmath342 is a representation embedding . in particular if @xmath34 then @xmath10 is of wild cm type .
by construction we have the commutative diagram of functors : @xmath343 \ar^-\phi[d ] & { \mathbf{mcm}}_r \ar^-{\pi}[d ] \\
{ \mathbf{ulr}}_t \ar^-{{\underline{\boldsymbol{\omega}}}_r^c}[r ] & { \underline{\mathbf{mcm}}}_r } \ ] ] we proved in theorem [ stable ff ] that @xmath344 is fully faithful , and in theorem [ strictly wild by extensions ] that @xmath124 is also fully faithful .
so the same happens to @xmath345 and therefore to @xmath346 .
therefore , if @xmath97 and @xmath129 are two representations of @xmath94 such that @xmath347 , we see that @xmath348 and hence @xmath349 by full faithfulness .
moreover , if @xmath350 is decomposable , then @xmath351 contains a non - trivial idempotent @xmath272 .
the class @xmath269 is also an idempotent , which is trivial if and only if the direct summand of @xmath350 associated with @xmath272 is free . but
this can not happen by lemma [ not free ] , again by full faithfulness of @xmath346 , @xmath269 corresponds to a non - trivial idempotent of @xmath97 , so @xmath97 is also decomposable .
this finishes the proof that @xmath342 is a representation embedding .
the consequence that @xmath10 is of wild cm type is clear . the first class of varieties where it is a priori unknown how to construct large families of acm bundles is given by general cubic hypersurfaces of dimension @xmath352 .
we can do this with theorem [ embedding ]
. indeed , start with a cubic hypersurface @xmath10 in @xmath353 , sufficiently general to admit a smooth surface section @xmath26
. then we may take @xmath354 and @xmath355 , where @xmath0 and @xmath356 are twisted cubics in @xmath26 meeting at @xmath357 points , cf .
@xcite : these will satisfy the assumptions of theorem [ embedding ] . in the next section
we will see how to deal in a similar fashion with any variety besides the non - wild varieties listed in the main result .
we prove here the main theorem of the introduction .
let @xmath358 be an integral closed acm subvariety of dimension @xmath186 and degree @xmath175 , and let @xmath26 be an integral one - dimensional linear section of @xmath10 .
we argue on the arithmetic genus @xmath48 of @xmath26 , i.e. the _ sectional genus _ of @xmath10 .
we also introduce the _ @xmath359-genus _ of @xmath10 , defined as @xmath360 ; it is well - known that @xmath361 . in case
@xmath51 we have @xmath362 , which means that @xmath10 has minimal degree .
according to the classification of del pezzo and bertini of these varieties , cf .
@xcite , we have that @xmath10 is smooth , for otherwise @xmath10 would be a cone . in this case
@xmath10 is of finite cm type ( cf .
@xcite ) if it is a linear space ( cf .
@xcite ) , or a smooth quadric ( see @xcite ) , or the veronese surface in @xmath15 , or a smooth cubic scroll in @xmath16 ( see @xcite ) , or a rational normal curve . also , @xmath10 is of tame cm type if it is a rational normal surface scroll of degree @xmath18 , cf .
@xcite . besides these cases
, @xmath10 is strictly ulrich wild , as we see by applying theorem [ strictly wild by extensions ] to the ulrich line bundles constructed in @xcite .
let us look at the case of ( arithmetic ) genus @xmath363 .
consider the normalization @xmath364 of @xmath26 , which is a smooth irreducible curve , let @xmath365 be the genus of @xmath364 and @xmath366 be the normalization .
we first consider an ulrich sheaf of rank one on @xmath26 . according to @xcite , to construct @xmath367 on @xmath26 we need to pick a sufficiently general line bundle @xmath368 of degree @xmath369 on @xmath364 and define @xmath370 .
since @xmath26 is integral , the sheaf @xmath367 is simple , being reflexive of rank @xmath20 .
now we consider two general flat deformations @xmath371 and @xmath372 of @xmath367 in the compactified jacobian of @xmath26 .
these are locally free sheaves of rank @xmath20 on @xmath26 , both having the same degree as @xmath367 , namely again @xmath369 . the condition that @xmath373 is ulrich amounts to @xmath374 for @xmath375 . by assumption
this vanishing holds for @xmath376 , hence by semicontinuity also for @xmath377 , so that @xmath371 and @xmath372 are ulrich line bundles on @xmath26 .
since the dimension of the compactified jacobian of @xmath26 is @xmath52 , we may assume that @xmath371 and @xmath372 are not isomorphic .
therefore , since the @xmath378 are invertible , are not isomorphic and have the same hilbert polynomial , by irreducibility of @xmath26 we deduce @xmath379 for @xmath380
. next , since @xmath371 and @xmath372 are locally free of rank @xmath20 , we have @xmath381 for @xmath382 .
therefore , we may choose @xmath29 and @xmath30 as two sheaves corresponding to non - trivial elements of @xmath383 , and write : @xmath384 it is clear that @xmath29 and @xmath30 are locally free ulrich sheaves of rank @xmath49 .
also it is easy to check that @xmath29 and @xmath30 are simple and satisfy : @xmath385 we obtain the following : @xmath386 note that the non vanishing condition of theorem [ stable ff ] reduces to @xmath387 , which is obviously true .
now theorem [ embedding ] implies that @xmath10 is of wild representation type .
finally , let us consider the case of sectional genus @xmath54 .
since @xmath10 is acm , it turns out that the @xmath359-genus of @xmath10 is also one , cf .
@xcite , namely @xmath10 is of almost minimal degree .
according to fujita s terminology , @xmath10 is a del pezzo variety .
this time , @xmath10 is of tame cm type if it is a smooth elliptic curve ( see @xcite ) or a rational curve of arithmetic genus @xmath20 with only an ordinary double point ( see @xcite ) , which settles the case @xmath388 . for higher dimension
, we have to work with a surface section @xmath26 of @xmath10 instead of a curve section .
so we let @xmath26 be a general linear section of @xmath10 of dimension @xmath49 , in particular @xmath26 is an integral projective acm surface which is not a cone , having degree @xmath175 in @xmath389 .
these surfaces constitute a generalization of classical del pezzo surfaces to allow certain singularities .
they are arithmetically gorenstein , as if fact @xmath390 . in particular
@xmath26 satisfies the non vanishing condition of theorem [ stable ff ] , namely @xmath391 .
surfaces of almost minimal degree and actually even higher dimensional varieties of almost minimal degree , are completely classified cf .
@xcite for the case of normal varieties , and @xcite for the non - normal case .
we will actually rely on the complete classification of ( singular ) del pezzo surfaces of degree @xmath392 and @xmath18 .
we refer to @xcite for this theory , rooted in work of schlfli and cayley , see for instance @xcite . for our purpose
, we have to study in two different ways the case of normal surfaces and that of non - normal ones .
indeed , in the normal case we can treat ulrich bundles of rank @xmath49 in view of the good knowledge of sets of points in @xmath26 and hartshorne - serre correspondence , while we may not rely on rank-@xmath20 ulrich bundles ( one reason is that they do not even exist for cubic surfaces with an @xmath393 singularity , cf .
* theorem 9.3.6 ) ) . on the other hand for the non - normal case we do use rank-@xmath20 ulrich sheaves coming from the normalization @xmath364 , which are under control since @xmath364 is a surface of minimal degree .
we assume from now on in this subsection that _ @xmath26 is ( locally ) normal_. this case is treated via the hartshorne - serre construction in the spirit for instance of @xcite , where we only need to modify the argument slightly so that it applies to singular surfaces too .
so let us consider a set @xmath394 of @xmath395 points in general linear position .
we have : @xmath396 where we used that @xmath397 and that @xmath185 spans the whole @xmath389 . through serre correspondence , a non - zero element @xmath398 provides a coherent sheaf @xmath75 of rank @xmath49 that fits into the short exact sequence : @xmath399 given a set of points @xmath185 of @xmath395 points of @xmath26 , write @xmath400 $ ] for the corresponding element of the hilbert scheme @xmath401 of subschemes of length @xmath395 of @xmath26 . in the next lines , stability is always with respect to @xmath402 .
a. [ li ] if @xmath185 is a reduced set of @xmath395 points in general linear position then @xmath404 is ulrich . b. [ lii ] if moreover @xmath185 is disjoint from the singular locus of @xmath26 then @xmath404 is locally free .
c. [ liii ] if @xmath400 $ ] is general enough in @xmath401 , then the sheaf @xmath404 is stable .
the fact that @xmath404 is ulrich follows from the form of the minimal graded free resolution of the ideal of @xmath185 , which can be extracted from @xcite , so is clear .
the fact that @xmath404 is locally free is immediate when @xmath26 is smooth , by the auslander - buchsbaum formula .
if @xmath26 is singular and @xmath185 does not meet the singular locus of @xmath26 , then @xmath185 is locally a complete intersection . on the other hand
, we have : @xmath405 so for any @xmath406 , the element @xmath407 represents a non - trivial extension of @xmath408 .
moreover , taking dual of the short exact sequence : @xmath409 we deduce : @xmath410 and therefore @xmath411 is generated by @xmath412
. then we can apply ( * ? ? ?
* lemma 5.1.2 ) to conclude that @xmath408 is free .
we have proved .
the proof of will be done by a dimension count .
if the ulrich bundle @xmath404 is not stable , a straightforward modification of @xcite , relying on the fact that @xmath413 , shows that @xmath404 is an extension @xmath414 where @xmath29 and @xmath30 are ulrich sheaves of rank @xmath20 . by construction , the global section @xmath415 associated with @xmath185 vanishes in codimension two .
therefore @xmath416 can not lie in @xmath417 , as @xmath416 would then then vanish in codimension @xmath20 .
therefore , we construct the following commutative diagram : @xmath418 \ar[d]^s & { \mathscr{o}}_y \ar[d ] & \\ 0 \ar[r ] & { \mathscr{a}}\ar[r ] \ar@{=}[d ] & { \mathscr{e}}_z \ar[r ] \ar[d ] & { \mathscr{b}}\ar[r]\ar[d ] & 0\\ 0 \ar[r ] & { \mathscr{a}}\ar[r ] & { \mathscr{i}}_{z|y}(2 ) \ar[r ] & { \mathscr{t}}\ar[r ] & 0 } \ ] ] where @xmath419 is a torsion sheaf defined by the diagram .
this says that : @xmath420 namely @xmath185 lies on a divisor @xmath1 from the linear system @xmath421 . since @xmath422 is also an ulrich sheaf ( again because @xmath413 ) , the dimension of the linear system @xmath421 is @xmath423 .
therefore , the family of subschemes of @xmath26 consisting of @xmath424-tuples of points lying on a divisor of class @xmath422 is at most : @xmath425 then , @xmath404 is stable as soon as @xmath400 $ ] lies away from the union of proper closed algebraic subsets of @xmath401 , one for each ulrich sheaf of rank @xmath20 over @xmath26 .
now the statement is clear since all of these sheaves have discrete parameter space .
indeed , any such sheaf corresponds to the linear equivalence class of a weil divisor of @xmath26 , and since @xmath26 is a normal regular surface , the divisor class group of @xmath26 is discrete . this construction can be performed in quite a general setup , for instance for cubic surfaces one even knows , if @xmath8 is not algebraically closed , the degree of the field extension needed to construct @xmath426 , cf .
however what we need is the following result .
the first two statements concerning morphisms are clear since we can suppose that @xmath11 and @xmath75 are stable with the same slope . for the first statement , since @xmath11 is locally free , we have @xmath428 . tensoring the short exact sequence [ serre ] by @xmath429 and considering the associated long exact sequence of global sections , since @xmath430
, we get an isomorphism @xmath431 on the other hand , the exact sequence defining @xmath432 twisted by @xmath433 reads : @xmath434 taking into account that @xmath435 we obtain @xmath436 .
then , tensoring by @xmath429 taking global sections and combining with we get : @xmath437 now we know that @xmath438 has @xmath439 independent global sections , as @xmath11 is ulrich . on the other hand , since @xmath185 has length @xmath395 , so that @xmath440 is just a vector space of rank @xmath49 concentrated at @xmath395 points , hence @xmath441 . we conclude that @xmath442 .
we may argue that this proposition says that any surface @xmath26 of almost minimal degree supports infinitely many ulrich bundles of rank @xmath49 with determinant @xmath433 , in particular any cubic surface admits ( over an algebraically closed field ) infinitely many pfaffian representations .
the normal case is now complete by virtue of theorem [ embedding ] .
this time @xmath26 is a non - normal acm surface of degree @xmath175 in @xmath389 , we let @xmath443 be the normalization of @xmath26 and @xmath444 the normalization map . it turns out that these surfaces are completely classified cf .
@xcite , and that @xmath364 is a rational normal scroll or a veronese surface . also , the locus where @xmath445 is not an isomorphism is a line @xmath446 and @xmath447 is a conic in @xmath443 . in this subsection
we are going to treat the case when @xmath443 is a rational normal scroll leaving the case of the veronese surface for the next subsection .
so let us consider the two non - isomorphic rank-1 ulrich bundles over @xmath443 , normalized so that they have @xmath175 independent global section , where @xmath448 is the degree of @xmath443 as a rational normal scroll , which is the same as the degree of @xmath26 .
these line bundles are @xmath449 , where @xmath274 is a _
fibre _ of the scroll map @xmath450 , i.e. @xmath274 is just a line of the scroll , and @xmath451 , where @xmath320 is the hyperplane section of @xmath443 .
let us write @xmath452 and @xmath453 .
we have : @xmath454 therefore , in order to conclude by theorem [ embedding ] , we only need to prove that in the cases @xmath460 either @xmath461 has dimension at least @xmath392 or that @xmath462 has dimension at least @xmath463 .
luckily , these two cases are completely classified up to projective equivalence , cf .
@xcite .
we said that @xmath26 is not normal along a line @xmath43 .
we have : @xmath464 in order to see this notice that the restriction of @xmath445 is @xmath465 as a map @xmath466 .
therefore @xmath467 is locally free of rank @xmath49 over @xmath43 , and @xmath468 , so we get an exact sequence : @xmath469 where @xmath470 is a line bundle on @xmath43 . from the previous exact sequence we have @xmath471 , namely @xmath472 .
next we may compute a locally free resolution of @xmath474 ( in fact the sheafified minimal graded free resolution of the module of global sections of @xmath474 ) over @xmath26 , relying on the classification appearing in @xcite .
this reads , when @xmath25 : @xmath475 and for @xmath22 : @xmath476 a general global section of @xmath480 vanishes along a rational normal curve @xmath356 of degree @xmath423 in @xmath443 , while any non - zero global section of @xmath481 vanishes along the union @xmath0 of @xmath423 fibres , which are pairwise disjoint lines contained in @xmath443 .
all this is clear from the description of divisors on @xmath443 , independently on the characteristic of @xmath8 .
it turns out that the zero locus of a general global section of @xmath29 is a curve @xmath482 which is the union of @xmath483 and @xmath43 , and the same happens to @xmath30 with obvious notation .
we write @xmath484 and @xmath485 , so @xmath486 and @xmath487 .
actually @xmath488 cuts @xmath43 at a single point @xmath489 , while @xmath490 cuts @xmath43 at @xmath423 points @xmath491 , one for each component of @xmath490 .
both @xmath482 and @xmath492 are acm curves in @xmath26 .
when @xmath356 is smooth ( which happens if the defining section of @xmath480 is general enough ) , setting @xmath48 ( resp .
@xmath493 ) for a point of @xmath356 ( resp . of @xmath488 ) , we get : @xmath494 we also have : @xmath495 there is an exact commutative diagram : @xmath496 & { \mathscr{o}}_y \ar[d ] \ar[r ] & { \mathscr{a}}\ar[r ] \ar@{=}[d ] & { \mathscr{a}}|_{\ha } \ar[d]\ar[r ] & 0 \\ 0 \ar[r ] & \sigma_*{\mathscr{o}}_\ty \ar[r ] & { \mathscr{a}}\ar[r ] & { \mathscr{a}}|_{a ' } \ar[r ] & 0 \\ } \ ] ] this induces : @xmath497 first
, we replace each instance of @xmath0 in the diagram by @xmath356 and we apply @xmath500 to the first row of the resulting exact commutative diagram .
since @xmath29 is locally cm we get an isomorphism : @xmath501 we first note that easily implies @xmath509 since @xmath490 and @xmath356 meet transversely along @xmath488 , i.e. away from the non - normal locus @xmath43 . in turn , this sheaf is obviously isomorphic to @xmath510 .
so we would like to check that @xmath510 has length @xmath423 by showing that it has rank @xmath20 over each of the points @xmath491 of @xmath511 .
looking at each point @xmath512 we have to show that the line @xmath513 defined as the irreducible component of @xmath490 passing through @xmath512 satisfies @xmath514 .
the situation is local , and the cases @xmath22 and @xmath25 do not differ in this sense , so we just look at @xmath22 .
we consider the sheafified minimal graded free resolution of the module of global sections of @xmath515 , which in turn looks precisely like with @xmath43 replaced with @xmath516 .
of course , @xmath517 , and we see that @xmath518 , where @xmath519 is the residual conic of @xmath516 with respect to a plane section of @xmath26 containing @xmath516 .
write @xmath520 for the restriction of @xmath521 to @xmath522 .
next we observe that , since @xmath516 meets @xmath43 transversely at @xmath512 , the sheaf @xmath523 is just the dual over @xmath43 of @xmath524 , that is @xmath525 .
the curve @xmath519 also meets @xmath43 transversely at @xmath512 , so : @xmath526 to continue our computation , we observe that the sheaf @xmath499 is locally free of rank @xmath49 over @xmath43 .
this follows from , and since @xmath531 and @xmath532 locally behave like @xmath533 , so @xmath30 and @xmath29 locally behave like @xmath534 .
then @xmath535 for @xmath536 , because we have just seen that the middle sheaf is locally free over @xmath43 in view of and the leftmost sheaf has finite length .
it remains to compute @xmath539 .
observe that in order to finish we have to show @xmath540 . to get this
, we apply @xmath541 to after replacing @xmath0 with @xmath356 . using
, it is easy to observe that : @xmath542 note that @xmath543 is supported at @xmath488 for all @xmath504 and @xmath544 is locally free over @xmath43 , so we see get an exact sequence : @xmath545 according to @xcite there are @xmath49 inequivalent integral non - normal cubic surfaces which are not cones in characteristic different from @xmath49 , and one more in characteristic @xmath49 ; the normalization of all of them is a smooth cubic scroll in @xmath16 . also , there are three inequivalent integral non - normal surfaces of degree @xmath18 in @xmath16 whose normalization is a quartic scroll and which are not cones , one of them appearing in characteristic @xmath49 .
all these cases can be treated with the method we outlined .
the computations sketched here can also be carried out explicitly with macaulay2 , @xcite , cf .
the webpage http://dfaenzi.perso.math.cnrs.fr/code/cm-wild we still have to study the case when @xmath443 is the veronese surface .
again there is a line @xmath446 which is the non - normal locus of @xmath26 . in this case , the two quadrics threefolds @xmath549 whose equations generate @xmath550 can be chosen so that @xmath551 is singular along a line @xmath552 .
the intersection @xmath553 consists a double point @xmath554 whose reduced structure @xmath104 lies in @xmath43 . also , @xmath555 is a cone over a point @xmath556 .
then , @xmath551 admits a unique linear symmetric determinantal representation by a @xmath557 matrix whose cokernel @xmath371 is a rank-1 ulrich sheaf on @xmath551 with @xmath49 independent global sections , corresponding to a pencil of planes @xmath558 in @xmath551 . on the other hand
@xmath555 supports two non - isomorphic rank-1 ulrich sheaves @xmath372 and @xmath559 , one for each of the two distinct pencils of planes @xmath560 and @xmath561 in @xmath555 .
we define : @xmath562 moreover , @xmath29 and @xmath30 are not isomorphic , as one can see by comparing the curves associated with them .
indeed , the curves associated with @xmath29 corresponding to decomposable tensors in @xmath563 are reducible of the form @xmath564 , and these curves can not appear as zero locus of global sections in @xmath565 , as @xmath561 would lie in the same pencil as @xmath560 .
it follows that : @xmath566 as any morphism @xmath567 would be an isomorphism , since the source and target are torsionfree of rank @xmath20 with the same hilbert polynomial , and the variety @xmath26 is integral .
it remains to compute @xmath568 .
we would like to show that this vector space has dimension @xmath392 , which suffices to conclude by theorem [ embedding ] .
note that @xmath371 is locally free away from @xmath70 , while @xmath372 and @xmath559 are locally free except at @xmath48 , so that @xmath29 is locally free away from @xmath48 and @xmath104 .
the same happens to @xmath30 , in particular @xmath499 is supported ( set - theoretically ) at @xmath48 and @xmath104 so it has finite length . by our analysis of reducible curves appearing as zero loci of sections of decomposable tensors in @xmath29 , we see that the only singularity of such a curve is , when @xmath558 and @xmath560 are general enough , the point @xmath569 .
therefore a general global section of @xmath29 will vanish on a smooth irreducible curve , actually a rational normal quartic @xmath0 in @xmath26 , and one has : @xmath570 where @xmath571 and @xmath572 are invertible on @xmath0 and @xmath356 .
any such curve @xmath0 or @xmath356 contains @xmath554 and @xmath48 . also
, two curves @xmath0 and @xmath356 corresponding to sufficiently general global sections of @xmath29 and @xmath30 intersect at the length-@xmath392 subscheme consisting of the fixed points @xmath554 and @xmath48 , and moreover at @xmath392 more simple points @xmath573 which move along with the choice of @xmath0 and @xmath356 .
this says that @xmath574 has length @xmath575 and is supported at @xmath554 , @xmath48 , @xmath576 , @xmath577 , @xmath578 .
recall that @xmath30 is ulrich and has thus @xmath18 independent global sections . on the other hand
, one immediately computes @xmath582 and @xmath583 , while of course @xmath584 .
so applying @xmath585 to the second sequence of one gets : @xmath586 using these sequences and we finally obtain @xmath587 . in characteristic @xmath49 , there is another surface @xmath26 of degree @xmath18 , not projectively equivalent to @xmath26 , whose normalization is the veronese surface , cf .
case 8 of @xcite .
however this case can be treated in the same way as the present one , so we omit the details .
the hypothesis that @xmath10 is integral and not a cone can be removed for sectional genus at least @xmath392 thanks to the following remark . in order to state it ,
we will use the notion of @xmath588-vector , cf .
@xcite for a thorough exposition . given an acm closed subscheme @xmath19 of dimension @xmath186 , the hilbert function of a linear artinian reduction of @xmath10
is eventually zero and therefore it is encoded on a finite sequence of positive integers @xmath589 .
this sequence is known as the _ h - vector _ of @xmath10 .
its length is @xmath590 .
the sectional genus @xmath48 of @xmath10 can be recovered by the following formula : @xmath591 set @xmath593 $ ] , so that @xmath68 is a graded cm ring .
we have @xmath594 so either @xmath595 or @xmath596 ; assume @xmath595 .
let @xmath597 be a homogeneous element of degree @xmath598 and @xmath599 .
then the hilbert series @xmath600}$ ] and @xmath601}$ ] of @xmath6 $ ] and @xmath28 $ ] are related as follows : @xmath602}(t)=(1-t^a)h_{{\boldsymbol k}[x]}(t).\ ] ] suppose now @xmath388 .
let @xmath603 ( respectively , @xmath604 ) be a @xmath68-regular sequences with @xmath605 ( respectively @xmath606 ) .
consider then @xmath607/ ( f_1,f_2)$ ] and @xmath608/(g_1,g_2)$ ] .
the hilbert series of @xmath609 and @xmath101 correspond , respectively , to the standard @xmath588-vector of @xmath10 and to the weighed one with weight @xmath610 .
now , the hilbert functions @xmath611 and @xmath612 of @xmath609 and of @xmath101 are related by : @xmath613 in fact , we have a more detailed information : if @xmath617 , then the resolution of @xmath618/(f_1,f_2)$ ] over @xmath619/(g_1,g_2)$ ] reads : @xmath620 then one easily gets the already mentioned relation between the hilbert functions of @xmath609 and @xmath101 .
we thank aldo conca for pointing out to us this fact .
maurice auslander and idun reiten .
almost split sequences for @xmath621-graded rings . in _
singularities , representation of algebras , and vector bundles ( lambrecht , 1985 ) _ , volume 1273 of _ lecture notes in math .
_ , pages 232243 .
springer , berlin , 1987 .
david eisenbud and joe harris . on varieties of minimal degree ( a centennial account ) . in _ algebraic geometry , bowdoin , 1985 ( brunswick , maine , 1985 ) _ , volume 46 of _ proc .
pure math .
_ , pages 313 .
soc . , providence , ri , 1987 .
|
we show that all projective integral varieties which are not cones and have a cohen - macaulay coordinate ring are of wild cm type , except for a few cases which we completely classify .
|
it has recently become clear that galaxy discs evolve significantly over time due to both internal and external influences ( see * ? ? ?
* ; * ? ? ?
* for reviews ) . scattering by giant molecular clouds @xcite is now thought to play a minor role @xcite , while spirals @xcite and interactions @xcite are believed to cause more substantial changes . the milky way is believed to have both a thin and a thick disc @xcite , as do perhaps many disc galaxies @xcite .
relative to the thin disc , the thick disc has a higher velocity dispersion , lags in its net rotational velocity @xcite , contains older stars with lower metallicities @xcite and enhanced [ @xmath0/fe ] ratios @xcite .
a number of different criteria have been used to divide stars into the two disc populations , and some of these trends may depend somewhat on whether a spatial , kinematic , or chemical abundance criterion is applied @xcite . recently , @xcite suggest that there is no sharp distinction into two populations , but rather a continuous variation in these properties .
several models have been proposed for the formation of the thick disc : accretion of disrupted satellite galaxies @xcite , thickening of an early thin disc by a minor merger @xcite , star formation triggered during galaxy assembly @xcite , and radial migration @xcite . while it is likely that no single mechanism plays a unique role in its formation , our focus in this paper is on radial migration .
as first shown by @xcite , spiral activity in a disc causes stars to diffuse radially over time .
stars near corotation of a spiral pattern experience large angular momentum changes of either sign that move them to new radii without adding random motion . as spirals are recurring transient disturbances that have corotation radii scattered over a wide swath of the disc ,
the net effect is that stars execute a random walk in radius with a step size ranging up to @xmath1kpc , which results in strong radial mixing .
note that this mechanism is distinct from the `` blurring '' caused by epicycle oscillations of stars about their home radii , @xmath2 , where @xmath3 in the midplane and @xmath4 is the gravitational potential .
radial mixing is dubbed `` churning '' and causes the home radii themselves to change .
note that radial mixing caused by recurrent transient spirals has two related properties that distinguish it from other processes that have sometimes also been said to cause radial migration .
the two distinctive properties are the absence of disc heating , already noted , and the fact that stars mostly change places , causing little change to the distribution of angular momentum within the disc and leaving the surface density profile unchanged . these are both features of scattering at corotation ; spirals also cause smaller angular momentum changes at lindblad resonances which do heat and spread the disc .
bar formation causes greater angular momentum changes than those of an individual spiral pattern @xcite .
however , because the bar persists after its formation , the associated angular momentum changes have a different character from those caused by a spiral that grows and decays quickly .
bar formation changes the radial distribution of angular momentum , and therefore also the surface density profile of the disc , as well as adding a substantial amount of radial motion @xcite .
although the bar may subsequently settle , grow , and/or slow down over time , the main changes associated with its formation probably occur only once in a galaxy s lifetime ( for a dissenting view see * ? ? ?
* ) whereas spirals seem to recur indefinitely as long as gas and star formation sustain the responsiveness of the disc .
thus angular momentum changes in the outer disc beyond the bar should ultimately be dominated by the effects of radial mixing by dynamically - unrelated transient spirals , as we find in this paper .
note that @xcite studied the separate physical process of resonance overlap using steadily rotating potentials to represent a bar and spiral , and therefore did not address possible mixing by transient spirals in a barred galaxy model .
both @xcite and @xcite find that external bombardment of the disc can enhance radial motion through the increase in the central attraction and also through possible angular momentum changes .
however , the importance of bombardment is unclear since satellites may be dissolved by tidal shocking @xcite before they settle , and furthermore satellite accretion to the inner milky way over the past @xmath5 years is strongly constrained by the dominant old age of thick disc stars @xcite .
@xcite discovered spiral churning in 2d simulations , and it has been shown to occur in fully 3d simulations @xcite .
@xcite present a model for galactic chemical evolution that includes radial mixing , and point out @xcite that it naturally gives rise to both a thin and a thick disc , under the assumption that thick disc stars experience a similar radial churning .
@xcite show that extensive spiral activity in their simulations causes a thick disc to develop , and present a detailed comparison with data @xcite from sdss @xcite .
@xcite and @xcite also generated galaxies with realistic break radii in the exponential surface brightness profiles using cosmological @xmath6-body hydrodynamic simulations that included radial migration .
however , @xcite report that no significant thick discs developed in their simulations .
@xcite find evidence for radial mixing in the thin disc and for a downtrend of rotational velocity with metallicity .
the presence of such a trend does not rule out migration in the thick disc , but shows that mixing for the oldest stars can not be complete .
@xcite find a continuous change of abundance - dependent disc structure with increasing scale height and decreasing scale length which they note is the `` almost inevitable consequence of radial migration '' .
the decreasing metallicity gradient with age of milky way disc stars can also be attributed to radial mixing @xcite . while @xcite present circumstantial evidence for radial migration in the thick disc , they do not show explicitly that it is occurring , neither do they attempt to quantify the extent to which it may be reduced by the weakened responses of thick disc stars to spiral waves in the thin disc .
@xcite find that mixing is more extensive when spiral activity is invigorated by star formation , although the level of spiral activity is strongly dependent on the `` gastrophysical '' prescription adopted .
they show that mixing persists even for particles with large oscillations about the mid - plane , and they determine migration probabilities from their simulations . in this paper , we set ourselves the limited goal of determining the extent of radial migration in isolated , collisionless discs with various thicknesses and radial velocity dispersions that are subject to transient spiral perturbations .
following @xcite , we first present controlled simulations of two - component discs constructed so as to support an isolated , large - amplitude spiral wave , in order to study the detailed mechanism of migration in the separate thin and thick discs .
we also report the responses of two - component discs to multiple spiral patterns , both with and without a bar .
all our models use the constant velocity disc known as the mestel disc ( * ? ? ?
* hereafter bt08 ) .
linear stability analysis by ( * ? ? ?
* see also zang 1976 and evans & read 1998 ) revealed that this disc with moderate random motion lacks any global instabilities whatsoever when half its mass is held rigid and the centre of the remainder is cut out with a sufficiently gentle taper .
is large . ]
we therefore adopt this stable model for the thin disc , and superimpose a thick disc of active particles that has @xmath7 of the mass of the thin disc .
the remaining mass is in the form of a rigid halo , set up to ensure that the total central attraction in the midplane is that of the razor - thin , full - mass , untapered mestel disc .
ideally , we would like to select particles from a distribution function ( ) for a thickened mestel disc .
@xcite found a family of flattened models that are generalizations of the razor - thin zang discs that have the isothermal @xmath8 vertical density profile @xcite .
unfortunately , these two - integral models have equal velocity dispersions in the radial direction and normal to the disc plane ( see bt08 ) , whereas we would like to set up models with flattened velocity ellipsoids .
since no three - integral for a realistic disc galaxy model is known , as far as we are aware , we start from the two - integral for the razor - thin disc @xcite : @xmath9 where @xmath10 is a particle s specific energy and @xmath11 its specific @xmath12-angular momentum .
the free parameter @xmath13 is related to the nominal radial component of the velocity dispersion through @xmath14 with @xmath15 being the circular orbital speed at all radii .
the value of toomre s local stability parameter for a single component , razor - thin mestel disc would be @xmath16 where @xmath17 is the active fraction of mass in the component ; note that these expressions for both @xmath18 and @xmath19 are independent of radius .
a composite model having two thickened discs , such as we employ here , will have some effective @xmath19 that is not so easily expressed .
we choose different values of @xmath18 for the thin and thick discs , given in table [ tab : simulations ] below , in order that the thick disc has a greater radial velocity dispersion , as observed for the milky way .
we limit the radial extent of the discs by inner and outer tapers @xmath20[1+(l_z / l_o)^6]},\ ] ] where @xmath21 and @xmath22 are the central angular momentum values of the inner and outer tapers respectively .
the exponents are chosen so as not to provoke instabilities ( toomre , private communication ) .
we employ the same taper function for both discs and choose @xmath23 .
we further restrict the extent of the disc by eliminating all particles whose orbits would take them beyond @xmath24 , where @xmath25 is the central radius of the inner taper .
this additional truncation is sufficiently far out that the active mass density is already substantially reduced by the outer taper . and
is broadened by epicycle excursions .
the blue line of slope @xmath26 indicates the adopted exponential profile of the thin disc.,width=317 ] we thicken these disc models by giving each the scaled vertical density profile @xmath27 where @xmath28 with @xmath29 independent of @xmath30 .
while both this function and the usual @xmath8 function have harmonic cores , we prefer eq .
( [ zprof ] ) because it approaches @xmath31 more rapidly , as suggested by data @xcite ; @xmath29 is therefore the exponential scale height when @xmath32 .
we set the vertical scale height of the thick disc to be three times that of the thin disc , as suggested for the milky way @xcite .
we estimate the equilibrium vertical velocity dispersion at each @xmath12-height by integrating the 1d jeans equation ( bt08 , eq .
4.271 ) in our numerically - determined potential .
this procedure is adequate when the radial velocity dispersion is low , but the vertical balance degrades in populations having larger initial radial motions , which flare outwards as the model relaxes from initial conditions , as shown below .
the tapered thin mestel disc has a surface density that declines with radius as shown in fig .
[ fig : run94surfdn ] .
while there is no radial range that is closely exponential , we estimate an approximately equivalent exponential radial scale length @xmath33 as shown by the blue line .
accordingly , we choose @xmath34 for the thin disc so that the ratio @xmath35 is similar to that of the milky way and the average volume - corrected ratio of @xmath36(@xmath37 ) found by @xcite for 34 nearby galaxies .
the values we adopt for each model are given in table [ tab : simulations ] below .
the radial gravitational potential gradient in the midplane of our model is less than that of the full - mass , razor - thin , infinite mestel disc as a result of the reduced disc surface density , the angular momentum tapers , the finite thickness of the discs , and gravitational force softening . in order to create an approximate equilibrium
, we therefore add a rigid central attraction to the self - consistent forces from the particles in the discs at each step .
we tabulate the supplementary central attraction needed at the initial moment to yield a radial force per unit mass of @xmath38 in the midplane , and apply this unchanging extra term as a spherically symmetric , rigid central attraction . .
the bottom left panel shows the initial @xmath39 of the thin ( bottom curves ) and thick ( top curves ) discs ( solid black ) compared to that at @xmath40 ( dashed red).,width=317 ] 1 [ cols="<,^,<,>,>,>,>,>,>,^",options="header " , ] the biweight estimated standard deviation of the fractional change in variable @xmath41 listed in the third column for both the thin and thick discs in simulations m2 and t. we give two values for the standard deviation in some rows : the value in the fourth column is measured from all the particles , that in the fifth column is from only the 48% ( in m2 ) of particles for which @xmath42 is calculable .
the second rows in table [ tab : ezjzbiweight ] show the rms fractional changes in the instantaneous values of @xmath43 . while these values are slightly smaller than for the epicyclic estimate , they are again large for the thin disc and still greater for the thick .
this is hardly surprising , as the particles in the simulation have random orbit phases at the two measured times .
particles on eccentric orbits generally spend more time at radii @xmath44 than inside this radius , which biases the instantaneous measure to a lower value , as shown in the third panel of fig . [
fig : run94orbvenrg ] . to eliminate this bias , and to reduce the random variations , we evaluate @xmath45 from eq .
( [ ezr ] ) at the home radius of each particle , which requires us to integrate the motion of each particle in the frozen potential of the appropriate time until the particle reaches its home radius .
a tiny fraction , @xmath46 of the @xmath47 sample particles , never cross @xmath48 at the initial time and @xmath49 more at the final time ; these particles rise to large heights above the mid - plane and have meridional orbits resembling that shown in the top right of fig .
[ fig : run94vactorbts ] , but appear to be confined to @xmath50 . the fact that this is possible seems consistent with the description of vertical oscillations developed by @xcite .
the third rows of table [ tab : ezjzbiweight ] give the fractional rms changes in @xmath45 for all the remaining particles ; the changes in simulation m2 are a little smaller than those of the instantaneous values and significantly so in simulation t. since @xmath43 is multiply periodic ( fig .
[ fig : run94orbvenrg ] ) , we have also estimated an orbit - averaged value , @xmath51 , by integrating the motion for many periods until the time average changed by @xmath52 when the integration is extended for an additional radial period .
the fourth rows of table [ tab : ezjzbiweight ] give the rms changes in @xmath51 , which are still considerable .
note that we did not compute this time - consuming estimate for all the particles , but for only the subset used for other values in the fifth column of this table .
the choice of this subset is described below .
generally , we find that none of these estimates of the vertical energy is even approximately conserved , except for the orbit - averaged energy when the simulation is constrained to remain axisymmetric ( simulation t ) . in this case
, the potential at the two times differs slightly due to radial variations in the mass distribution
values of this orbit - averaged quantity in an unchanged potential would be independent of the moment at which the integration begins .
( first row ) , @xmath43 ( second row ) , @xmath45 ( third row ) , and @xmath53 ( fourth row ) versus @xmath54 for the thin ( left column ) and thick ( right column ) discs of simulation m2 . just as in fig .
[ fig : run94zmaxchlch ] , the horizontal and vertical red lines show zero changes , the linearly spaced magenta contours show number density , and the bold solid green and dashed blue curves show the mean and median changes in each ordinate respectively .
note that the vertical scales differ in each plot.,width=317 ] in all cases , changes are larger in simulation m2 where significant radial migration occurs .
[ fig : run94ezchlch ] illustrates that the changes in the estimated vertical energy correlate with @xmath55 .
the excesses of particles in the second and fourth quadrants indicate that changes have a tendency to be negative for outwards migrating particles and positive for inwards migrating particles .
the significance of this trend is discussed below .
the various actions of a regular orbit in a steady potential are defined to be @xmath56 times the appropriate cross - sectional area of the orbit torus ( bt08 , pp .
211215 ) .
one advantage of actions is that they are the conserved quantities of an orbit when the potential changes slowly they are adiabatic invariants under conditions that are defined more carefully in bt08 ( pp . 237238 ) . for changes in @xmath57 ( first row ) , @xmath58 ( second row ) , and @xmath59 ( third row ) of the particles in m2.,width=317 ]
the vertical action is defined as @xmath60 where @xmath12 and @xmath61 are measured in some suitable plane that intersects the orbit torus .
since @xmath62 is conserved in a steady axisymmetric potential , the orbit can be followed in the meridional plane ( bt08 , p159 ) in which it oscillates both radially and vertically , as illustrated in the first panel of fig .
[ fig : run94orbvenrg ] . to estimate @xmath59
, we need to integrate the orbit of a particle in a simulation from its current position in the frozen , azimuthally averaged potential at that instant and construct the @xmath63 surface of section ( sos ) as the particle crosses @xmath2 with @xmath64 , say . the integral in eq .
( [ vertact ] ) is the area enclosed by an invariant curve in this plane . before embarking on this elaborate procedure ,
we consider two possible approximations . when the epicycle approximation holds , the vertical action is @xmath65 ( bt08 p. 232 ) .
the fifth rows of table [ tab : ezjzbiweight ] for each disc show that changes in this quantity are large and again they are similar to those in simulation t , confirming once again that the epicycle approximation is inadequate .
since most orbits reach beyond the harmonic region ( @xmath66 ) of the vertical potential , an improved local approximation is to calculate @xmath67 at the particle s fixed position in the disc .
this local estimate still ignores the particle s radial motion , but gives a useful estimate for the average vertical action that is also used in analytic disc modeling @xcite .
the function @xmath68 at this fixed point is simply determined by the vertical variation of @xmath69 , and the area is easily found .
we evaluate @xmath58 at the particle s instantaneous position at the initial and final times using the corresponding azimuthally averaged frozen potential .
the rms variation of @xmath70 is given in the sixth rows of table [ tab : ezjzbiweight ] for each disc .
we find that @xmath70 in the thin discs of both m2 and t are small , suggesting that this estimate of vertical action is more nearly conserved .
however , the changes for the thick discs are still large , and we conclude that this local estimate is still too approximate .
we therefore turn to an exact evaluation of eq .
( [ vertact ] ) using the procedure described in the appendix .
unfortunately , we can evaluate @xmath42 only if the consequents in the sos form a simple invariant curve that allows @xmath59 to be estimated at both times .
we find that only 48% of the @xmath47 particles have closed , concave invariant curves at the initial time and only 73% of those retain these properties at the final time .
although the thick disc contains more particles than the thin , we are able to calculate @xmath42 for a smaller fraction : we succeed with @xmath71 in the thin disc but only @xmath72 in the thick .
column five of table [ tab : ezjzbiweight ] gives the rms changes of all estimates of vertical energy and action for only these particles .
while the fractional rms values of the different energy estimates remain large , they are generally smaller than those for the all particles given in the fourth column , but only slightly so for @xmath58 .
thus the orbits for which @xmath42 can be computed are not a random subset , but are biased to those for which energy changes are smaller on average . for @xmath73 particles from simulation m2 at the initial and final times .
values are calculated from eq .
( [ vertact ] ) .
the top panel is on a linear scale , the bottom on a log scale to reveal the behaviour for very small actions .
the red line has unit slope .
thin disc particles are marked in black , thick disc in green.,width=317 ] the seventh ( final ) rows of table [ tab : ezjzbiweight ] for each disc list the rms values of the fractional change in @xmath42 .
they are just a few percent for particles in simulation t , in which the disc is constrained to remain axisymmetric , and are smaller than for any other tabulated quantity in the discs perturbed by a spiral .
the small scatter about the red line of unit slope in fig .
[ fig : run94vact ] indicates that differences in the final and initial values of @xmath59 in both discs for m2 are indeed small .
[ fig : run94jzchlch ] shows that changes in all three estimators of the vertical actions are uncorrelated with @xmath54 , and that the mean and median changes are close to zero .
comparison with the systematic changes in fig .
[ fig : run94ezchlch ] suggest that @xmath59 is , in fact , conserved . to see this , note that changes in @xmath11 shift the home radius of each particle to a region where the average vertical restoring force differs
, the amplitudes of the vertical oscillations increase due to the weaker average restoring force when @xmath74 , and conversely decrease for @xmath74 .
thus if an estimator of vertical motion is not conserved , we should expect a systematic variation with @xmath54 , as we observe for all the energy estimators in fig .
[ fig : run94ezchlch ] .
the fact that there is no systematic variation of the median or mean change in vertical action suggest that it is conserved .
if @xmath59 is a conserved quantity , it may seem puzzling that the rms changes in our measurements are not smaller . both @xmath75 and @xmath58 are approximate , which could be responsible for apparent substantial changes , but changes in the `` exact '' estimate of @xmath59 remain significant , even for simulation t. it is possible that our estimate of @xmath42 is inaccurate .
for example , we must eliminate non - axisymmetric structure from the potential to compute @xmath59 , introducing small errors when the potential is mildly non - axisymmetric .
larger differences could arise , however , if the invariant curve changes its character , especially since entering or leaving a trapped area in phase space is not an adiabatic change .
it is possible an orbit that appears to be unaffected by resonant islands in phase space at the initial and final times could have experienced trapping about resonant islands for some of the intermediate evolution , allowing @xmath59 to change even when changes to the potential are indeed slow .
we have verified that @xmath59 is conserved to a part in @xmath76 in a further simulation with a frozen , axisymmetric potential .
note this is not a totally trivial test , since we use the @xmath6-body integrator and the grid - determined accelerations to advance the motion for @xmath77 dynamical times before making our second estimate .
therefore the non - zero changes in simulation t are indeed caused by changes to the potential . even though the potential variations are small
, substantial action changes indicate that changes to the orbit were non - adiabatic , which can happen for the reason given in the previous paragraph .
our `` initial '' @xmath59 values are estimated at @xmath78 , when we believe the model has relaxed from the initial set up ; we therefore suspect that the small potential variations are more probably driven by particle noise , which can be enhanced by collective modes @xcite .
this hypothesis is supported by our finding variations in @xmath59 that were about ten times greater than in simulation t when we employed 100 times fewer particles .
we have presented a quantitative study of the extent of radial migration in both thin and thick discs in response to a single spiral wave .
we find angular momentum changes in the thick disc are generally smaller than those in the thin , although the tail to the largest changes in each population is almost equally extensive .
we have introduced populations of test particles into a number of our simulations in order to determine how changes in @xmath11 vary with disc thickness and with radial velocity dispersion when subject to the same spiral wave , finding an exponential decrease in @xmath79 , as shown in fig .
[ fig : run102103rmslch ] .
we find that spirals of smaller spatial scale cause smaller changes .
when we were careful to change all the properties of the model by the appropriate factors , we were able to account for smaller changes to @xmath79 as being due to a combination of the weakened spiral amplitude near corotation and the change in the value of @xmath80 . furthermore , we found evidence that the saturation amplitude scales inversely as @xmath80 , in line with the theory developed by @xcite .
note the simple scaling holds true only when the principal dynamical properties , such as @xmath19 , @xmath81 , thickness , and gravity softening , are held fixed relative to the scale of the mode .
nevertheless , it seems reasonable to expect smaller changes to @xmath79 in general for spirals higher @xmath80 .
the exponential decrease of @xmath79 with increasing disc thickness applies only for different populations subject to the same spiral perturbation .
we have also run slightly more realistic simulations to follow the extent of churning in both thick and thin discs that are subject to large numbers of transient spiral waves having a variety of rotational symmetries . fig .
[ fig : run104rhfrhi ] shows that changes in the home radii of thick disc particles are smaller on average than those of the thin disc , but again the tails of the distributions in both populations are almost co - extensive .
as found in previous work @xcite , the formation of a bar also causes substantial angular momentum changes within a disc , but we find that the churning effect from multiple spiral patterns still dominates changes in the outer disc after the bar has formed . we find that vertical action is conserved during radial migration , despite the fact that relative changes in our estimated @xmath59 , which can be measured for only about half the particles , are as large as @xmath82 . because the vertical restoring force to the mid - plane decreases outwards , an increased ( decreased ) home radius causes the particle to experience a systematically weaker ( stronger ) vertical restoring force , making it impossible for both vertical energy and action to be conserved .
we find a clear systematic variation of the vertical energy with @xmath54 , but none with @xmath42 leading us to conclude that vertical action is the conserved quantity .
the residual scatter in our measured values of @xmath42 could be caused by trapping and escape from multiply - periodic resonant islands in phase space , as well as numerical jitter in the @xmath6-body potential , as we discuss at the end of 6 . in the absence of these complications , we believe that the vertical action is conserved
. thus conservation of vertical action , and not of vertical energy , should be used to prescribe the changes to vertical motion in models of chemo - dynamic evolution .
it would be especially useful to obtain diffusion coefficients that include the effects of radial migration as functions of radius and height due to a transient spiral or a combination of consecutive spirals with various corotation radii .
we leave this work to a future paper .
we thank the referee , rok roskar , for a thorough and helpful report .
this work was supported in part by nsf grant ast-1108977 to jas .
abadi , m. g. , navarro , j. f. , steinmetz , m. & eke , v. r. 2003 , , * 597 * , 21 beers , t. c. , flynn , k. & gebhardt , k. 1990 , , * 100 * , 32 bensby , t. , feltzing , s. , lundstrm , i. & ilyin , i. 2005 , , * 433 * , 185 binney , j. 2010 , , * 401 * , 2318 binney , j. j. & lacey , c. g. 1988 , , * 230 * , 597 binney , j. & tremaine , s. 2008 , galactic dynamics .
princeton univ . press , princeton ( bt08 ) bird , j.
c. , kazantzidis , s. & weinberg , d. h. 2011 , , submitted ( arxiv:1104.0933 ) bournaud , f. & combes , f. 2002 , , * 392 * , 83 bournaud , f. , elmegreen , b. g. & martig , m. 2009 , , * 707 * , l1 bovy , j. , rix , h - w . , liu , c. , hogg , d. w. , beers , t. c. & lee , y. s. 2011 , arxiv:1111.1724 brook , c. b. , gibson , b. k. , martel , h. & kawata , d. 2005 , , * 630 * , 298 brunetti , m. , chiappini , c. & pfenniger , d. 2011 , , * 534 * , a75 camm , g. l. 1950 , , * 110 * , 305 chiba , m. & beers , t. c. 2000 , , * 119 * , 2843 debattista , v. p. , mayer , l. , carollo , c. m. , moore , b. , wadsley , j. & quinn , t. 2006 , , * 645 * , 209 donghia , e. , springel , v. , hernquist , l. & keres , d. 2010 , , * 709 * , 1138 evans , n. w. & read , j. c. a. 1998 , , * 300 * , 106 friedli , d. , benz , w. & kennicutt , r. 1994 , , * 430 * , l105 fuhrmann , k. 2008 , , 173 gilmore , g. & reid , n. 1983 , , * 202 * , 1025 grenon , m. 1999 , , * 265 * , 331 hnninen , j. & flynn , c. 2002 , , * 337 * , 731 helmi , a. , navarro , j. f. , nordstrm , b. , holmberg , j. , abadi , m. g. & steinmetz , m. 2006 , , * 365 * , 1309 hohl , f. 1971 , , * 168 * , 343 ivezi ' c , .
2008 , , * 684 * , 287 juri ' c , m. 2008 , , * 673 * , 864 kazantzidis , s. , bullock , j. s. , zentner , a. r. , kravtsov , a. v. & moustakas , l. a. 2008 , , * 688 * , 254 kregel , m. , van der kruit , p. c. & de grijs , r. 2002 , , * 334 * , 646 lee , y. s. 2011 , , * 738 * , 187 loebman , s. r. , roskar , r. , debattista , v. p. , ivezi , z. , quinn , t. r. & wadsley , j. 2011 , , * 737 * , 8 majewski , s. r. 1993 , , * 31 * , 575 martnez - serrano , f. j. , serna , a. , domnech - moral , m. & domnguez - tenreiro , r. 2009 , , * 705 * , l133 mestel , l. 1963 , , * 126 * , 553 monaghan , j. j. 1992 , , * 30 * , 543 minchev , i. & famaey , b. 2010 , , * 722 * , 112 minchev , i. , famaey , b. , combes , f. , di matteo , p. , mouhcine , m. & wozniak , h. 2011 , , * 527 * , a147 munn , j. a. 2004 , , * 127 * , 3034 quillen , a. c. , minchev , i. , bland - hawthorn , j. & haywood , m. 2009 , , * 397 * , 1599 quinn , p. j. , hernquist , l. & fullagar , d. p. 1993
, , * 403 * , 74 raboud , d. , grenon , m. , martinet , l. , fux , r. & udry , s. 1998 , , * 335 * , l61 reddy , b. e. , lambert , d. l. & allende prieto , c. 2006 , , * 367 * , 1329 roskar , r. , debattista , v. p. , quinn , t. r. , stinson , g. s. & wadsley , j. 2008a , , * 684 * , l79 roskar , r. , debattista , v. p. , stinson , g. s. , quinn , t. r. , kaufmann , t. & wadsley , j. 2008b , , * 675 * , l65 snchez - blzquez , p. , courty , s. , gibson , b. k. & brook , c. b. 2009 , , * 398 * , 591 scannapieco , c. , white , s. d. m , springel , v. & tissera , p. b. 2011 , , * 417 * , 154 schnrich , r. & binney , j. 2009 , , * 396 * , 203 schnrich , r. & binney , j. 2009 , , * 399 * , 1145 schnrich , r. & binney , j. 2012 , , * 419 * , 1546 sellwood , j. a. 1985 , , * 217 * , 127 sellwood , j. a. 2010 , in _ planets stars and stellar systems _ ,
v.*5 * , ed .
g. gilmore ( heidelberg : springer ) to appear ( arxiv:1006.4855 ) sellwood , j. a. 2012 , , submitted sellwood , j. a. & athanassoula , e. 1986 , , * 221 * , 195 sellwood , j. a. & binney , j. j. 2002 , , * 336 * , 785 sellwood , j. a. & carlberg , r. g. 1984 , , * 282 * , 61 sellwood , j. a. & debattista , v. p. 2009
, , * 398 * , 1279 sellwood , j. a. & kahn , f. d. 1991 , , * 250 * , 278 sellwood , j. a. & preto , m. 2002 , in disks of galaxies .
dynamics and perturbations . ,
, eds . e. athanassoula & a. bosma a. ( san francisco ) sellwood , j. a. & valluri , m. 1997 , , * 287 * , 124 shen , j. & sellwood , j. a. 2004 , , * 604 * , 614 spitzer , l. 1942 , , * 95 * , 329 spitzer , l. & schwarzschild , m. 1953 , , * 118 * , 106 toomre , a. 1977 , , * 15 * , 437 toomre , a. 1981 , in the structure and evolution of normal galaxies , eds .
s. m. fall & d. lynden - bell ( cambridge , cambridge univ .
press ) p. 111
toomre , a. 1982 , , * 259 * , 535 van der kruit , p. c. & freeman , k. c 2011 , , * 49 * , 301 villalobos , . & helmi a. 2008 , , * 391 * , 1806 weinberg , m. d. 1998 , , * 297 * , 101 wyse , r. f. g. 2009 , in `` the galaxy disk in cosmological context '' , iau symposium * 254*. eds
. j. andersen , j. bland - hawthorn & b. nordstrm ( cambridge , cambridge university press ) p. 179
( arxiv:0809.4516 ) yoachim , p. & dalcanton , j. j. 2006 , , * 131 * , 226 york , d. g. , 2000 , , * 120 * , 1579 yu , j - c . , sellwood , j. a. , pryor , c. hou , j - l . & li , c. 2012 , , ( submitted ) zang , t. a. 1976 , , mit
note that we show only a small fraction of the full orbit used to produce the invariant curves in the @xmath83 plane .
the vertical red lines mark @xmath2 at which the sos is constructed .
the initial vertical position and radial and vertical velocities of each of these three particles are noted.,width=317 ] fig .
[ fig : run94vactlecnst ] illustrates the sos for five different energies , all for @xmath84 .
each invariant curve is generated by a test particle in the initial potential of m2 .
all the particles in one panel have the same energy and angular momentum ( noted in the top right corner ) , but differ in the extent of their vertical motion .
the zero - velocity curve , where a particle s radial velocity must be zero for the @xmath11 and @xmath10 values adopted is the outer most green curve .
we see that most of phase space is regular , but not all consequents mark out simple closed invariant curves ; some significant fraction of phase space is affected by islands in the sos that surround multiply periodic orbits .
chaotic motion , in which consequents fill an area rather than lie on a closed curve in the sos , is extensive only for the greatest energy .
the paths in the meridional plane of three multiply periodic orbits are shown in fig .
[ fig : run94vactorbts ] .
they correspond to the bold red invariant curves in the second , fourth , and fifth sos panels of fig .
[ fig : run94vactlecnst ] .
the last orbit has almost equal radial and vertical periods and circulates only clockwise in the meridional plane .
since we plot only when the particle passes @xmath2 with @xmath85 , the consequents lie only in the first quadrant of the @xmath83 plane . were we to plot the other crossing instead , where @xmath86 , the invariant curve would be a reflected image in the second quadrant .
also , had we chosen the negative root for initial radial velocity , the particle would have circulated only counterclockwise and yielded a @xmath87-rotationally symmetric invariant curve in the third quadrant .
this accounts for the skewness of the invariant curves in the sos , which diminishes for particles of lower energy .
the area in the sos enclosed by only those orbits having a simple invariant curve yields a numerical estimate of the vertical action @xmath59 .
we find all the consequents for a particle as its motion advances over @xmath88 dynamical times from its position in the frozen , azimuthally averaged , potential , and make a numerical estimate of the area integral ( eq .
[ vertact ] ) .
|
we present a study of the extent to which the sellwood & binney radial migration of stars is affected by their vertical motion about the midplane .
we use both controlled simulations in which only a single spiral mode is excited , as well as slightly more realistic cases with multiple spiral patterns and a bar .
we find that rms angular momentum changes are reduced by vertical motion , but rather gradually , and the maximum changes are almost as large for thick disc stars as for those in a thin disc .
we find that particles in simulations in which a bar forms suffer slightly larger angular momentum changes than in comparable cases with no bar , but the cumulative effect of multiple spiral events still dominates .
we have determined that vertical action , and not vertical energy , is conserved on average during radial migration .
[ firstpage ] galaxies : kinematics and dynamics galaxies : evolution galaxies : structure .
|
unitary operations have been placed in a very important position in the quantum communication and entanglement manipulating , such as , quantum cryptography@xcite , teleportation@xcite , entanglement swapping@xcite , quantum states purification[4 ] , entanglement production@xcite and so on . in quantum teleportation , to transfer the unknown quantum state to the remote user , the sender must apply an joint unitary operator on the unknown state particle and one of the entangled particles . in quantum entanglement swapping , a joint unitary transformation on two particles(they
are from two different entangled pairs ) will let two remote particles entangled without direct interaction . in entanglement purification process , joint unitary operations and measurements can transfer the entanglement from many partially entangled pairs to few near perfect entangled pairs . in entanglement generation , the joint unitary operations and single qubit operations can let the initially product particles entangled . from the above applications we can see that , it is the nonlocal attribute of the bipartite joint unitary transformation that plays the most important role .
the nonlocal attribute of a bipartite joint unitary operator has been studied from different aspects , such as entangling power@xcite , operator entanglement@xcite , and entanglement - changing power@xcite .
entangling power is the mean entanglement(linear entropy ) produced by acting with u on a given distribution of pure product states@xcite . because a quantum operator belongs to a hilbert - schmidt space
, one can consider the entanglement of the operator itself , which is named as operator entanglement@xcite .
it is a natural extension of the entanglement measures of quantum states@xcite to the level of general quantum evolutions . up to now
, several methods have been proposed to quantify the entanglement of an unitary bipartite operator , such as linear entropy@xcite , von neumann entropy@xcite , concurrence@xcite and schmidt strength@xcite etc .
the relations between the entangling power and these operator entanglement measures have also been discussed recently@xcite .
in general , the entangling power of an unitary operator is related to those operator entanglement measures in complicated or indirect ways , so does the relation between the different operator entanglement measures .
clarifying the exact relation between the entangling power and different operator entanglement measures , and the relation between different operator entanglement measures will be very helpful for us to understand the nonlocal attributes and entanglement capacity of a joint unitary operator .
after getting the whole nonlocal features of joint unitary operators , we can choose the optimal unitary operator to produce the specific entangled state as we want it to be , and the quantum communication protocols(such as teleportation , entanglement swapping etc ) can be realized in an optimal way by introducing the appropriate joint operations@xcite . for two - qubit unitary operators ,
the two operator entanglement measures schmidt strength and linear entropy are shown to have a one - to - one relation between them for the schmidt number @xmath0 case , but no such relation exists for the schmidt number @xmath1 case@xcite .
this result also shows that , the schmidt number is a very important parameter of an unitary operator when the entangling power and operator entanglement of it is concerned . in this paper
, we are going to study the operator entanglement of joint two - qubit unitary operators with different schmidt numbers . in this paper
, we use the linear entropy as the entanglement measure of joint two - qubit unitary operator@xcite , and study the schmidt number and the entanglement measure of any unitary operator in four - dimensional hilbert - schmidt space .
the schmidt number of two - qubit unitary operators has the following three possible situations : @xmath2 , @xmath0 or @xmath1@xcite .
we will show that the entanglement measure of two - qubit unitary operators is classified by the schmidt number of the unitary operator . in the light of the numerical analysis
, we can get the extreme value of the operator entanglement for the two - qubit unitary operators .
further , the relation between the operator entanglement and the parameters of the unitary operator will be clarified here .
there exist local unitary operators @xmath3,@xmath4 , @xmath5 , @xmath6 and a two - qubit unitary operator @xmath7 , so that arbitrary two - qubit unitary operator @xmath8 can be canonically decomposed as@xcite : @xmath9 where @xmath10 $ ] , and @xmath11 is a diagonal matrix . in the light of this theory ,
any bipartite unitary operator can be decomposed as the form above .
moreover , the entanglement measure of a unitary operator s must be invariant under the local unitary transformations@xcite .
so , the entanglement measure of any bipartite unitary operator can be simplified into the entanglement measure of operator @xmath7 . in the standard computational basis , we have@xcite : @xmath12 where @xmath13 , @xmath14 , and one can always restrict oneself to the region @xmath15 , which is the so - called weyl chamber@xcite .
any operator @xmath16 acting on the systems @xmath17 and @xmath18 can be written in the operator - schmidt decomposition@xcite : @xmath19 where @xmath20 are the schmidt coefficients with the positive value and @xmath21 , @xmath22 are orthonormal operator bases for @xmath17 and @xmath18 , respectively . to calculate the operator entanglement of the unitary operator @xmath8
, we only need to make the schmidt decomposition of the unitary operator @xmath23 . from the ref.@xcite , entanglement measure of a unitary operator
can be expressed as : @xmath24 where @xmath25 and @xmath26 are dimensions of @xmath17 and @xmath18 , respectively .
so we can get the entanglement measure for the unitary operator @xmath23 as follows : @xmath27\sin^{2}({c}_{1}+{c}_{2})\sin^{2}({c}_{1 } -{c}_{2})\\ \nonumber & & + [ 1 + 2\cos^{2}(2{c}_{3})]\cos^{2}({c}_{1}+{c}_{2})\cos^{2}({c}_{1}-{c}_{2})\}. \label{eud}\end{aligned}\ ] ] the schmidt number@xcite is the number of non - zero coefficients @xmath28 . for the unitary operator @xmath7 , the schmidt coefficients @xmath28 are as follows : @xmath29^{1/2},\ ] ] @xmath30^{1/2},\ ] ] @xmath31^{1/2},\ ] ] @xmath32^{1/2}.\ ] ] [ schmidtcoes ] we made numerical analysis for the schmidt number and the entanglement measure of the unitary operator , and got the relation between the schmidt number and the entanglement measure of the unitary operator(shown in table .
[ table1 ] ) .
.the schmidt number versus the entanglement measure of the unitary operator @xmath23 [ table1 ] . [ cols="<,>",options="header " , ] for the schmidt number @xmath1 case , the first plot in fig.[fig1 ] shows how the entanglement measure of @xmath7 depends on the parameters @xmath33 and @xmath34 for @xmath35 when the schmidt number of @xmath7 is @xmath1 . as the parameters @xmath33 and @xmath34 approach to @xmath36 , which represents a unit matrix , the entanglement measure of the unitary operator approaches to @xmath36 . as the parameters @xmath33 and @xmath34 are equal to @xmath37 , which represents the swap gate , the entanglement measure of the unitary operator is equal to the maximum value @xmath38 . as the parameters @xmath39 and @xmath40 are increasing , the entanglement measure of unitary operator @xmath7 is increasing too . if @xmath41 , the changing pattern of the operator entanglement is similar to that of the @xmath35 case . from the other three plots in fig.[fig1 ] , we can see that the minimum entanglement for the schmidt-@xmath1 operator is oscillating with @xmath42 , and the maximum value and the period of the oscillation are @xmath43 and @xmath44 , respectively . when @xmath45 , the minimum entanglement reaches its maximum @xmath43 .
the maximum value of operator entanglement is still @xmath38 .
entanglement measure of unitary operator @xmath7 versus the parameters @xmath33 , @xmath34 for parameter @xmath46 respectively , when the schmidt number is @xmath1 . ] for the schmidt number @xmath0 case , if @xmath47 , then @xmath48 must be zero , so the operator entanglement can be expressed in a very simple form @xmath49 .
fig.[fig2 ] shows how the entanglement measure of @xmath7 depends on the parameters @xmath33 and @xmath34 for @xmath35 when the schmidt number of @xmath7 is @xmath0 .
this curve is the boundary line of the first plot in fig.[fig1 ] . as the parameters @xmath33 and @xmath34 approach to @xmath36 , the entanglement measure of unitary operator @xmath7 approaches to @xmath36 .
the entanglement measure of unitary operator will increase as the parameters @xmath33 or @xmath34 increase .
the entanglement measure of the unitary operator with schmidt number @xmath0 can get to the extreme value @xmath50 .
entanglement measure of unitary operator @xmath7 versus the parameters @xmath33 , @xmath34 for parameter @xmath51 when the schmidt number is @xmath0 . ] from the two figures we can see that , if we want to design an operation so that it has a specific operator entanglement(or entangling power ) , we have infinite design schemes(i.e .
@xmath52 ) for the schmidt - number-@xmath1 type operations .
but we only have two design schemes for the schmidt - number-@xmath0 type operations .
that is to say , schmidt - number-@xmath1 type operations have a variety of design schemes rather than the only two design schemes of the schmidt - number-@xmath0 type operations .
so , the schmidt - number-@xmath1 type operations are superior to the schmidt - number-@xmath0 type operations .
in addition , the maximum operator entanglement of the schmidt - number-@xmath1 type operations can reach @xmath38 , while the maximum operator entanglement of the schmidt - number-@xmath0 type operations only reaches @xmath50 .
so , in practice , we will prefer the schmidt - number-@xmath1 type operations .
to demonstrate the abstract relation between the operator entanglement and the parameters of the unitary operator , we will take the following detailed example .
consider two two - level atoms(@xmath2 , @xmath0 ) trapped in a single - mode optical cavity , and the two atoms are coupled to the cavity mode with the same coupling constant @xmath53 .
the excited state@xmath54 and the ground state@xmath55 , @xmath56 are the two levels used to encode quantum information .
the two atoms have different transition frequencies @xmath57 , @xmath58 , and @xmath59 . the frequency of the cavity mode is denoted by @xmath60 .
the atom @xmath2 is resonantly driven by an external classical field with coupling constant @xmath61 .
suppose the cavity mode is initially prepared in vacuum state , under the large detuning condition @xmath62 , @xmath63 and in the strong driving regime @xmath64 , the effective hamiltonian of the total system can be expressed as@xcite : @xmath65 where @xmath66 is the effective coupling constant between atoms @xmath2 and @xmath0 , and @xmath67 is the pauli operator of the @xmath68th atom .
the unitary transformation induced by this effective hamiltonian can be expressed as : @xmath69 if we set @xmath70 , @xmath71 , @xmath51 in eq.([decomatrix ] ) , it is just the joint unitary operator in eq.([ueff ] ) .
that is to say , the above mentioned physical process is just a physical realization of the joint unitary operation ( [ decomatrix ] ) .
the schmidt number of the operator ( [ ueff ] ) is @xmath0 , and the operator entanglement measure of it can be expressed as @xmath72 .
the relationship between the operator s entanglement measure and the effective interaction time @xmath73 between the two atoms is depicted in fig.[fig3 ] . from this figure
we can easily see that the maximum operator entanglement is @xmath50 with @xmath74 .
entanglement measure of unitary operator @xmath75 versus the effective interaction time @xmath73 between the two atoms . here
the schmidt number of @xmath75 is @xmath0 . ]
in this paper , the linear entropy and the schmidt number of an arbitrary two - qubit unitary operator is discussed .
the results have shown that the schmidt number is related with the entanglement measure of unitary operators closely .
for the same operator entanglement within the range @xmath76 $ ] , there exist infinite unitary operators with schmidt number @xmath1 but only @xmath0 unitary operators with schmidt number @xmath0 . in this sense
, we can say that the unitary operators with schmidt number @xmath1 can be realized more easily than the unitary operators with schmidt number @xmath0 if the same operator entanglement is required .
in addition , for the unitary operators with schmidt number @xmath1 , the range for the operator entanglement is @xmath77 $ ] .
but , for the unitary operators with schmidt number @xmath0 , the range decline to @xmath76 $ ] . there must be some requirement of entanglement which can be available for the unitary operator with schmidt number @xmath1 only .
this work is supported by national natural science foundation of china ( nsfc ) under grants no .
10704001 , no . 61073048 , no.10905024 and 11005029 , the specialized research fund for the doctoral program of higher education(20113401110002 ) , the key project of chinese ministry of education.(no.210092 ) , the china postdoctoral science foundation under grant no .
20110490825 , anhui provincial natural science foundation under grants no .
11040606m16 and 10040606q51 , the key program of the education department of anhui province under grants no .
kj2012a020 , no .
kj2012a244 , no .
kj2010a287 , no .
kj2012b075 and no .
2010sqrl153zd , the ` 211 ' project of anhui university , the talent foundation of anhui university under grant no.33190019 , the personnel department of anhui province , and anhui key laboratory of information materials and devices ( anhui university ) .
99 n. gisin , g. ribordy , w. tittel , and h. zbinden , rev .
74 , 145(2002 ) . c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres , and w. k. wootters , phys
70 , 1895 ( 1993 ) .
m. zukowski , a. zeilinger , m. a. horne , and a. k. ekert , phys .
71 , 4287 ( 1993 ) . c. h. bennett , g. brassard , s. popescu , b. schumacher , j. a. smolin , and w. k. wootters , phys
76 , 722 ( 1996 ) .
a. g. white , d. f. v. james , p. h. eberhard , and p. g. kwiat , phys .
83 , 3103 ( 1999 ) .
p. zanardi , c. zalka , and l. faoro , phys .
rev . a 62 , 030301(r)(2000 ) .
p. zanardi , phys .
a 63 , 040304(r)(2001 ) .
wang , b. c. sanders , and d. w. berry .
a 67 , 042323 ( 2003 ) .
ye , d. sun , y .- s .
zhang , and g .- c .
guo , phys .
a 70 , 022326 ( 2004 ) .
m. a. nielsen and i. l. chuang . quantum computation and quantum information ( cambridge university press , cambridge , england,2000 ) .
s. bose , v. vedral , p. l. knight .
phys.rev . a 60 , 194 ( 1999 ) . w. k. wootters , phys .
80 , 2245 ( 1998 ) .
s. hill and w. k. wootters , phys .
78 , 5022 ( 1997 ) .
wang and p. zanardi , phys .
a 66 , 044303 ( 2002 ) .
m. a. nielsen , c. m. dawson , j. l. dodd , a. gilchrist , d. mortimer , t. j. osborne , m. j. bremner , a. w. harrow , a. hines , phys .
a 67 , 052301(2003 ) .
s. balakrishnan and r. sankaranarayanan , phys .
a 83 , 062320 ( 2011 ) .
the optimality of these quanutm communication protocols by introducing the appropriate joint unitary operations will be discussed in another work which will be published soon .
b. kraus and j. i. cirac .
a 63 , 062309 ( 2001 ) .
j. zhang , j. vala , s. sastry , k. b. whaley .
phys . rev .
a 67 , 042313 ( 2003 ) . a. t. rezakhani
a 70 , 052313 ( 2004 ) .
m. horodecki , p. horodecki , and r. horodecki .
phys . rev .
a 60 , 1888 ( 1999 ) .
jin , c .- s .
yu , p pei , and h .- s .
song , phys .
a 82 , 042112 ( 2010 ) .
|
operator entanglement of two - qubit joint unitary operations is revisited .
schmidt number is an important attribute of a two - qubit unitary operation , and may have connection with the entanglement measure of the unitary operator .
we found the entanglement measure of two - qubit unitary operators is classified by the schmidt number of the unitary operators .
the exact relation between the operator entanglement and the parameters of the unitary operator is clarified too .
|
graphics processing units ( gpus ) have been developed originally as co - processors meant to fast deal with graphics tasks . in recent years
the video game market developments compelled gpus manufacturers to increase the floating point calculation performance of their products , by far exceeding the performance of standard cpus in floating point calculations .
the architecture evolved toward programmable many - core chips that are designed to process in parallel massive amounts of data .
these developments suggested the possibility of using gpus in the field of high - performance computing ( hpc ) as low - cost substitutes of more traditional cpu - based architectures : nowadays such possibility is being fully exploited and gpus represent an ongoing breakthrough for many computationally demanding scientific fields , providing consistent computing resources at relatively low cost , also in terms of power consumption ( watts / flops ) .
due to their many - core architectures , with fast access to the on - board memory , gpus are ideally suited for numerical tasks allowing for data parallelism , for simd ( single instruction multiple data ) parallelization .
the numerical simulation , by monte carlo algorithms , of the path integral formulation of quantum field theories , discretized on a euclidean space - time lattice , is a typical such task : one has to sample the distribution for a system with many degrees of freedom and mostly local interactions .
quantum chromodynamics ( qcd ) , the quantum gauge theory describing strong interactions , is a typical example where numerical simulations represent the best tool to investigate systematically specific features of the theory and give an answer to many important unsolved questions , regarding e.g. color confinement , deconfinement and the values of hadron masses .
lattice qcd and its computational needs has represented a challenge in the field of hpc since more than 30 years , being often the stimulus for new machine developments ( think e.g. of the series of ape machines @xcite ) .
the introduction of gpus in lattice qcd calculations started with the seminal work of ref .
@xcite , in which the native graphics apis were used , but the real explosion of interest in the field followed the introduction of nvidia s cuda ( compute unified device architecture ) platform , that effectively disclosed the field of gpgpu ( general purpose gpu @xcite ) , providing a more friendly programming environment .
gpus have maintained their role of co - processors in most numerical applications , where they are used as accelerators for some specific , time demanding purposes . in the same spirit ,
most of previous studies on the application of gpus to lattice qcd calculations were mainly aimed at using them together with the standard architectures in order to speed up some specific steps , typically the expensive dirac matrix inversion .
our intent is instead to use gpus in substitution of the usual architectures , actually performing the whole simulation by them : one still needs a cpu to run the main program , but mostly in the role of a mere controller of the gpu instruction flow .
to achieve this result we found simpler to write a complete program from scratch instead of using existing software packages , in order to have a better control of all the steps to be performed and ultimately transferred to the gpu .
our implementation uses nvidia s cuda platform together with a standard c++ serial control program running on cpu .
the specific case considered in the present study regards qcd on a hypercubic lattice with quark fields discretized in the standard staggered formulation .
the paper is organized as follows . in section [ simalg ]
we give more details about the lattice discretization of qcd considered in our study and the simulation algorithm adopted . in section [ gpufeatures ] we review some of the fundamental features of gpu architectures . in section [ implem ]
we describe our implementation of the algorithm on gpus and discuss the achieved performances .
finally , in sections [ opencl ] and [ parallel ] we discuss some preliminary comparisons with performances obtained with opencl and multigpu implementations of our code .
a preliminary report about our implementation has been presented in ref .
qcd is a quantum field theory based on the symmetry under local non - abelian gauge ( color ) transformations belonging to the group of special unitary @xmath0 complex matrices , @xmath1 .
it describes six different flavors of spin 1/2 colored particles ( quarks ) , which transform in the fundamental ( triplet ) representation of @xmath1 and interact through the gauge field , which lives in the algebra of the color gauge group and describes 8 colored , spin 1 particles known as gluons .
an elegant , gauge invariant lattice discretization of the theory is given in terms of gauge link variables @xmath2 , where @xmath3 indicates a lattice site and @xmath4 is one of the four euclidean space - time directions @xcite .
they are the elementary parallel transporters belonging to the gauge group and associated to each elementary link connecting two neighboring sites of the lattice .
hence in total we have @xmath5 @xmath1 matrices , where @xmath6 is the number of lattice sites along direction @xmath4 , which in present simulations is typically not larger than @xmath7 .
quark fields @xmath8 instead live on lattice sites and carry a color index , hence they are complex color triplets , one for each flavor and dirac index .
the discretized euclidean feynman path integral , giving e.g. a representation of the thermal partition function , is written as @xmath9-\bar{\psi}m[u]\psi}\end{aligned}\ ] ] where @xmath10 is the pure gauge part of the action , describing gluon - gluon interactions and written in terms of traces of products of link variables over closed loops , while @xmath11\psi$ ] is a bilinear form in the fermionic variables , which describes quark - gluon interactions , with @xmath12 $ ] a large sparse matrix written in terms of the gauge link variables . the functional integration in eq .
( [ pathint ] ) is over all link variables ( gauge group invariant integration for each link ) and all quark fields .
actually , due to their fermionic nature , the quark fields appearing in the path integral are grassmann anticommuting variables ; the best way we know to numerically deal with them is to integrate them explicitly .
that results in the appearance of the determinant of the fermion matrix @xmath12 $ ] : @xmath13-\bar{\psi}m[u]\psi } \propto \int \mathscr{d}u \det(m[u ] ) e^{-s_g[u ] } \ ; .\end{aligned}\ ] ] notice that , in general , a fermion determinant appears for each quark species and that the determinant term becomes a trivial constant when all quarks have infinite mass and decouple ( pure gauge or quenched limit ) .
one can show that , apart from specific difficult cases ( e.g. qcd at finite baryon density ) , the integrand in eq .
( [ determinant ] ) is a positive quantity , admitting a probabilistic interpretation , so that one can approach the numerical computation of the path integral by importance sampling methods , which are typically based on dynamic monte carlo algorithms .
the most difficult , time consuming part in such algorithms consists in taking properly into account the fermion determinant @xmath14)$ ] .
the best available method is to introduce dummy bosonic complex fields @xmath15 , which come in the same number as the original fermionic variables and are known as pseudo - fermions @xcite : @xmath16 ) \right)^2 e^{-s_g[u ] } \propto \int \mathscr{d}u \mathscr{d}\phi \
, \exp\big(-s_g[u]-\phi^ * \big(m[u]^{\dag}m[u]\big)^{-1}\phi\big)\end{aligned}\ ] ] where we have explicitly considered the case of two identical quark species , as in the case of two light flavors ( @xmath17 and @xmath18 quarks ) with all other flavors decoupled .
the particular lattice discretization implemented in the present study considers a simple plaquette action for the pure gauge term , i.e. products of four gauge link variables around the elementary closed square loops ( plaquettes ) of the lattice , and a standard staggered discretization for the fermionic term .
that means that fermion fields living on lattice sites have only color indexes ( dirac indexes can be reconstructed afterwards combining fields living on different lattice sites ) , while the fermionic matrix reads as follows : @xmath19 = a m \delta_{n_1,n_2 } + { 1 \over 2 } \sum_{\nu=1}^{4}\eta_\nu ( n_1 ) \left ( u_\nu ( n_1 ) \delta_{n_1,n_2-\hat\nu } - u^{\dag}_\nu ( n_1-\hat\nu ) \delta_{n_1,n_2+\hat\nu } \right ) \:,\end{aligned}\ ] ] where @xmath20 and @xmath21 are 4-vectors with integer components labeling lattice sites , @xmath22 is an elementary lattice versor , @xmath23 are known as staggered fermion phases and @xmath24 is the lattice spacing .
color indexes are implicit in eq .
( [ stag ] ) ( the identity in color space is understood for the mass term proportional to @xmath25 ) .
the staggered discretization differs from other ( e.g. wilson ) fermion discretizations for the absence of the additional dirac index : that implies lighter algebra computations which have an effect both on the overall computational complexity and on the maximal performances , as we shall explain in details later on . a particular feature of the staggered fermion matrix in eq .
( [ stag ] ) is that it describes four flavors . when simulating a different number of flavors one has to use a trick known as rooting .
@xmath26 flavors of equal mass are described by the following partition function @xmath27 ) \right)^{n_f/4 } e^{-s_g[u ] } \ , .\end{aligned}\ ] ] a convenient algorithm to simulate the action in is the hybrid monte carlo @xcite ( hmc ) .
the idea is very simple and it is conveniently exposed by using as an example the case of a single variable with action @xmath28 , i.e. distributed proportionally to @xmath29 . as a first step a dummy variable @xmath30 , corresponding to a momentum conjugate to @xmath31 , is introduced using the following identity : @xmath32 it is trivial that expectation values over @xmath31 are untouched by the introduction of @xmath30 , which is a stochastically independent variable .
the basic idea of the hmc algorithm is to sample the distribution in @xmath30 and @xmath31 by first extracting a value of @xmath30 according to its gaussian distribution and then performing a joint molecular dynamics evolution of @xmath30 and @xmath31 which keeps the total `` energy '' @xmath33 unchanged , obtaining an updated value of @xmath31 as a final result . going into more details , the hmc proceeds as follows : 1 .
a random initial momentum is generated with probability @xmath34 ; 2 .
starting from the state @xmath35 , a new trial state @xmath36 is generated by numerically solving in the fictitious time @xmath37 the equations of motion derived from the action @xmath38 , @xmath39 such equations are integrated numerically using a finite time step @xmath40 . as a consequence
the energy is conserved only up to some power of @xmath40 , depending on the integration scheme adopted ; 3 .
the new state @xmath36 is accepted with probability @xmath41 where @xmath42 ( metropolis step ) .
it can be shown ( see @xcite ) that the sequence of the @xmath31 configurations obtained in this way is distributed with the correct @xmath43 probability provided the solution of the equation of motion satisfies the requirements * the evolution is reversible , @xmath44 * the evolution preserves the measure of the phase space , @xmath45 a large class of integrators that satisfy these two constraints are the so - called symmetric symplectic integrators , the simplest member of this class being the leap - frog or @xmath46 scheme ( for improved schemes see @xcite ) . in the particular case of the action in , the momenta are conjugate to the gauge link variables : they are therefore @xmath0 complex matrices @xmath47 ( one for each gauge link ) living in the algebra of the gauge group , i.e. they are traceless hermitian matrices writable as @xmath48 where @xmath49 are the group generators , and the action associated with momenta is simply given by @xmath50 . a convenient implementation of the picture above
is then given by the so - called @xmath51 algorithm of ref .
@xcite : 1 .
a vector @xmath52 of complex gaussian random numbers is generated and the pseudofermion field is initialized by @xmath53^{\dag } r$ ] , in such a way that the probability distribution for @xmath15 is proportional to @xmath54 ; 2 .
the momenta are initialized by gaussian random matrices ( i.e. each @xmath55 is extracted as a normally distributed variable ) ; 3 .
the gauge field and momenta are updated by using the equations of motion ; 4 .
the final value of the action is computed and the metropolis step performed .
point 3 is the more time consuming , since the calculation of the force requires at each step to solve the sparse linear system @xmath56^{\dag}m[u]\big)x=\phi\ ] ] which is usually performed by means of krylov methods ( see @xcite ) . for staggered fermions
a complication is the presence of the @xmath57th root of the determinant in the action : becomes @xmath58^{\dag}m[u]\big)^{n_f/4}x=\phi\ ] ] where @xmath26 is the number of degenerate flavors . in order to overcome this problem the rational hybrid monte carlo ( rhmc )
was introduced in @xcite , in which the root of the fermion matrix is approximated by a rational function , which is then efficiently computed by means of the shifted versions of the krylov solvers ( see @xcite ) . in order to speed - up the simulations , the following common tricks were implemented * even / odd preconditioning @xcite * multi - step integrator , with action divided in gauge and fermion part @xcite * improved integrator , second order minimum norm , see @xcite * multiple pseudo - fermions to reduce the fermion force magnitude and increase integration step size @xcite * different rational approximations and stopping residuals for the metropolis and the molecular dynamic steps @xcite
in this section we will review the main features of the gpus architecture which are to be taken into account in order to efficiently use their computational capabilities .
modern gpus are massively parallel computing elements , composed of hundreds of cores grouped into multiprocessors . the typical architecture of a modern nvidia graphic card is outlined in and the most important point for the following is the presence of three different storage levels . roughly speaking
the architecture of ati cards is similar .
primary storage is provided by the device memory , which is accessible by all multiprocessors but has a relatively high latency . within the same multiprocessor ,
cores have also access to local registers and to shared memory .
shared memory is accessible by the threads of the same multiprocessor and its access is orders of magnitude faster than device memory one , being very close to the computing units ; however , while the total amount of device memory is of order of few gbs , the local storage is only 16 kb per multiprocessor both for the registers and for the shared memory , so that it is typically impossible to use just these local fast memories . in order to hide the latency time of the device memory
it is convenient to have a large number of threads in concurrent execution , so that when data are needed from device memory for some threads , the ones ready to execute are immediately sent to computation .
the highest bandwidth from device memory is achieved when a group of 16 threads accesses a contiguous memory region ( coalesced memory access ) , because its execution requires just one instruction call , saving a lot of clock - cycles .
this will be crucial in the following , when discussing the storage model for the gauge configuration .
double precision capability was introduced with nvidia s gt200 generation , the first one specifically designed having in mind hpc market , and by now there is only a factor @xmath59 between the peak performance in single and double precision . in the specifications of the gpus used in this work
are reported .
.specifications of the nvidia cards used in this work .
[ cols= " < , < , < , < , < , < " , ] in general we have 3 different stages to synchronize the border : we build the buffer border from the field , we transfer it to the device , then we flush it into the field . in the particular case of parallelization along the t direction
we can reduce these steps to only the transfer one .
the build and flush stages add an overhead of about @xmath60 on big lattices ( comparable to a @xmath61 on the single gpu ) that reduces slowly when further increasing the lattice size . in order to hide the time needed for transfer , we try to overlap it with computation , and in particular with the shift update inside the inverter code .
regarding performances ( see table [ parallel_tab ] ) , on a @xmath62 we have an efficiency , compared to the single gpu case , of @xmath63 on two s2050 ( boost @xmath64 ) , of @xmath65 on two c1060 ( @xmath66 ) and of @xmath67 ( @xmath68 ) on four c1060 ; in the last case , increasing the lattice size , we can reach @xmath69 ( @xmath70 ) on a @xmath71 lattices .
all tests have been performed by splitting along the y direction and the inefficiencies can be explained by the use of the two additional kernels needed to align the border before communication . on large lattices
we obtain therefore a good scaling , comparable to what reported in ref . @xcite , which is promising for the extension to the multinode implementation at which we are currently working . on smaller lattices , instead , the transfer time can not be hidden anymore and the boost decreases rapidly .
we have a current new line of development to overcome the splitting problem , based on building the border according to the general topology of a given lattice operator , which will permit to compute the border part of an operator separately from the interior part , in order to overlap not only with the shift update but also with internal computations of the operator .
such improvement may be particularly useful in the case of improved multi - link actions and operators and may also introduce a better memory access pattern .
the extremely high computation capabilities of modern gpus make them attractive platforms for high - performance computations .
previous studies on lattice qcd applications have been devoted almost exclusively to the dirac matrix inversion problem .
we have shown that it is possible to use gpus to efficiently perform a complete simulation , without the need to rely on more traditional architectures : in this case the gpu is not just an accelerator , but the real computer .
our strategy therefore has been that of bringing as much as possible of the computations on the gpu , leaving for the cpu only light or control tasks : in particular the whole molecular dynamics evolution of gauge fields and momenta , which is the most costly part of the hybrid monte carlo algorithm , runs completely on the gpu , thus reducing the costly cpu - gpu communications at the minimum .
following such strategy , we have developed a single gpu code based on cuda and tested it on c1060 and c2050 architectures .
we have been able to reach boost factors up to @xmath72 as compared to what reached by a twin traditional c++ code running on a single cpu core .
our code is currently in use to study the properties of strong interactions at finite temperature and density and the nature of the deconfinement transition , in particular first production results have been reported in ref .
@xcite .
a point worth noting is that in our implementation we have to rely on the reliability of the gpu .
if the gpu is used just for the dirac matrix inversion the result can then be directly checked on the cpu without introducing significant overhead in the computation .
such a simple test can not be performed if the gpu is used to perform a complete md trajectory .
for this reason it was mandatory to use gpus of the tesla series . during the editorial processing of this paper
it was signaled us that also the twqcd collaboration uses gpus to completely perform the hybrid monte carlo update of qcd with optimal domain wall fermions ( their first results were published in @xcite ) . reported performances make surely gpus the preferred choice for medium size lattice groups who need enough computational power at a convenient cost , in this sense they already represent a breakthrough for the lattice community .
our current lines of development regard the extension of our code to opencl and to multigpu architectures and we have reported preliminary results about that in section [ further ] : that will open to possibility to use gpu clusters with fast connection links ( see for instance ref .
@xcite ) in order to make the gpu technology available also for large scale simulations .
it s a pleasure to thank a. di giacomo : without his encouragement and support much of the work reported here could not have been realized .
test simulations have been run mostly on two gpu farms located in pisa and genoa and provided by infn .
we thank massimo bernaschi , edmondo orlotti and the ape group in rome for the possibility of an early usage of fermi cards during the first stages of our test runs .
we thank t .- w .
chiu and m. a. clark for useful comments and one of the referees for his suggestion to improve the reversibility .
m. a. clark , a. d. kennedy _ accelerating dynamical fermion computations using the rational hybrid monte carlo ( rhmc ) algorithm with multiple pseudofermion fields_. phys .
lett . * 98 * , 051601 ( 2007 ) [ arxiv : hep - lat/0608015 ] .
m. a. clark , r. babich , k. barros , r. c. brower , c. rebbi _ solving lattice qcd systems of equations using mixed precision solvers on gpus_. comput .
commun . * 181 * , 1517 ( 2010 ) [ arxiv:0911.3191 [ hep - lat ] ] .
d. gddeke , r. strzodka , s. turek _ performance and accuracy of hardware - oriented native- , emulated- and mixed - precision solvers in fem simulations_. international journal of parallel , emergent and distributed systems * 22 * , 221 ( 2007 ) .
b. joo , b. pendleton , a. d. kennedy , a. c. irving , j. c. sexton , s. m. pickles , s. p. booth _ instability in the molecular dynamics step of hybrid monte carlo in dynamical fermion lattice qcd simulations_. phys .
d * 62 * , 114501 ( 2000 ) [ arxiv : hep - lat/0005023 ] .
chiu , t .- h .
hsieh , y .- y . mao ( for the twqcd collaboration ) _ topological susceptibility in two flavors lattice qcd with the optimal domain - wall fermion_. phys .
b * 702 * , 131 ( 2011 ) [ arxiv:1105.4414 ] .
r. ammendola , a. biagioni , o. frezza , f. l. cicero , a. lonardo , p. s. paolucci , d. rossetti , a. salamon , g. salina , f. simula , l. tosoratto , p. vicini _ apenet+ : high bandwidth 3d torus direct network for petaflops scale commodity clusters _ [ arxiv:1102.3796 ] .
|
we report on our implementation of the rhmc algorithm for the simulation of lattice qcd with two staggered flavors on graphics processing units , using the nvidia cuda programming language .
the main feature of our code is that the gpu is not used just as an accelerator , but instead the whole molecular dynamics trajectory is performed on it . after pointing out the main bottlenecks and how to circumvent them ,
we discuss the obtained performances .
we present some preliminary results regarding opencl and multigpu extensions of our code and discuss future perspectives . lattice qcd , graphics processing units
|
effective lagrangians @xcite have been used for a long time as a systematic method to incorporate the known symmetries of a problem into the quantum field theory language . however , with the advent of yang - mills theories and the higgs mechanism , which allowed the construction of physically interesting renormalizable theories , they were used only for those problems that could not be treated in any other way . particularly important have been its applications to low - energy strong interactions in the form of the so - called chiral perturbation theory @xcite and more recently the heavy quark effective field theory @xcite .
its application to weak interactions has been less intensive , since in the last decade the main effort has been in the direction of building complete renormalizable theories that could serve as alternatives or extensions of the standard model , by just enlarging the number of fermions , gauge bosons and scalars .
renormalizability was considered to be a main point and completely linked to predictability , because non - renormalizable models need an infinite number of parameters to be completely described .
the striking confirmation of many of the standard model predictions in lep experiments has started to change this point of view .
almost every one now thinks that the standard model correctly describes physics at present energies and perhaps also up to energies close to the tev range .
of course , supersymmetric particles and other elusive particles , such as neutral heavy leptons or right - handed neutrinos , with some hidden interactions and masses much lighter than 1 tev , are not excluded . on the other hand ,
the theoretical developments achieved since the beginning of the seventies have brought a new interpretation of renormalizability @xcite .
nowadays renormalizability is not understood as a calculational requirement or a consistency requirement .
in fact one knows how to calculate , for example , with the fermi lagrangian , as long as one does not try to use it beyond the fermi scale .
the key idea is that one does not expect a quantum field theory , even a renormalizable one , to be valid up to arbitrarily large energy scales .
renormalizability is then seen as the physical requirement that physics at low energies can not dramatically depend on the physics at some large scale
. there could be effects of the heavy particles on the low - energy physics , but all of them must be suppressed by some power of the scale @xmath1 at which the new physics starts .
then , for each range of energies one expects that physics is described by a lagrangian of the form : @xmath2 where @xmath3 is a renormalizable lagrangian that describes the low - energy physics and @xmath4 are linear combinations of non - renormalizable operators of dimension @xmath5 . for processes involving energies @xmath6 much smaller than @xmath1 ,
the effects of the non - renormalizable operators are suppressed by @xmath7 and can systematically be computed .
of course , they depend on more arbitrary parameters , but , if the underlying theory is known and if it is renormalizable , they can , in principle , be computed in terms of the few renormalizable couplings of the underlying theory by matching the calculation of several observables or green functions done with both theories : the full theory and the effective one . this picture is known as the decoupling case @xcite , as the effects of heavy particles decouple from the low - energy physics , and a well - known example is physics in the 580 gev range . there
, physics can be well described by a @xmath8-@xmath9 renormalizable gauge - invariant lagrangian involving the five light quarks and the leptons , supplemented by four - fermion non - renormalizable interactions that take into account the effects of weak interactions .
these non - renormalizable couplings can be computed in terms of the standard model couplings by matching some green functions at the fermi scale .
but once this matching is done all calculations , including @xmath8 and @xmath9 radiative corrections , can be performed at the effective lagrangian level . since the full standard model is @xmath8 and @xmath9 gauge invariant one obtains that all the effective couplings must also be @xmath8 and @xmath9 gauge invariant .
following the previous reasoning one can then expect that the standard model is not valid for an arbitrarily large range of energies and that it is just a low - energy approximation of a more complete theory .
there are many arguments suggesting that this is the case .
in particular , the so - called naturalness problem of the standard model , which will be discussed later on in the context of our effective lagrangian , and the large amount of arbitrary input parameters needed to describe the standard model are among the most popular ones .
if the standard model is just a low - energy approximation of a more complete theory , one would expect that new non - renormalizable couplings suppressed by the scale of the new physics should arise . in complete parallelism
one would also expect these non - renormalizable interactions to be gauge invariant with respect to the standard model gauge group .
however , a complication arises since the @xmath10 gauge symmetry of the standard model is spontaneously broken to @xmath11 .
then , when building an effective theory for physics beyond the standard model , there are two possibilities : * the gauge symmetry is linearly realized .
this means that radial excitations of the scalar field are relatively light ; then , for energies larger than the mass of these excitations , the symmetry is effectively restored .
this is the simplest situation , in which the effective lagrangian is constructed with exactly the same particle content as the standard model particle spectrum and non - renormalizable interactions are required to be @xmath10 gauge invariant .
* the gauge symmetry is non - linearly realized . that is , the physics of spontaneous symmetry breaking is non - linear .
the radial excitations are heavy with respect to the fermi scale ; then , there is a wide range of energies between the fermi scale and the scale of new physics in which the full standard gauge symmetry is present , but realized in a non - linear way . in this region of energies one can still write an effective lagrangian , but the first term of that lagrangian is a non - renormalizable non - linear sigma model @xcite .
there is a big difference between these two possibilities .
the second one assumes that the new physics is responsible for spontaneous symmetry breaking ; as a consequence it must start not far away from the fermi scale , since it must correct the bad behaviour of the standard model without higgs particles . in the first one ,
however , the higgs mechanism is fully implemented .
low - energy physics decouples completely from high - energy physics and the only way to get some hint on the scale of the possible new physics ( apart from naturalness arguments ) is just by looking at the size of the non - renormalizable interactions .
experimental bounds are then the only source of information .
a very simple and instructive illustration of the use of effective lagrangians to bound new physics is the so - called see - saw mechanism for the generation of neutrino masses @xcite .
it is not difficult to see that the only @xmath10 gauge - invariant operator of dimension five that can be built with the field content of the standard model is @xmath12 where @xmath13 is the standard left - handed doublet of leptons , @xmath14 , @xmath15 ( @xmath16 is the charge conjugation operator ) , @xmath17 is the higgs doublet and @xmath18 ; @xmath19 is a complex symmetric matrix in flavour space ( @xmath20 and flavour indices have been suppressed ) .
it is clear that this lagrangian does not conserve generational lepton numbers , but in addition it does not conserve the total lepton number , which is violated in two units .
this kind of operator will be generated in any theory that does not conserve lepton number .
when the higgs develops a vacuum expectation value ( vev ) , it will generate a neutrino majorana mass matrix given by @xmath21 if we take the largest eigenvalue of @xmath19 to be of order 1 , the higgs vev @xmath22= 174 gev and use the experimental bound on the @xmath23-neutrino mass , @xmath24 mev , we find that @xmath25 gev
. should one take the cosmological bound @xcite on neutrino masses , @xmath26 ev , one would obtain @xmath27 gev .
these bounds are really impressive . but
what is @xmath1 ?
the lagrangian of eq .
( [ see - saw1 ] ) seems to be generated by the exchange of a scalar triplet with hypercharge 1 between the leptons and the higgses .
then , @xmath1 should be the mass of that triplet .
however this is not the only possibility .
in fact , eq . ( [ see - saw1 ] ) can be identically rewritten , after a @xmath20 fierz transformation , as @xmath28 which suggests the exchange of a neutral heavy majorana fermion ; then @xmath1 should be the mass of that fermion .
indeed , the original formulation of the see - saw mechanism@xcite was based on this possibility .
this simple example shows the power and the limitations of the effective lagrangian approach .
one can set impressive bounds on the scale of new physics , but one can not completely disentangle its origin , at least by taking into account only the lowest - order operators . it must also be remarked that in the effective lagrangian approach it is essential to know what the low - energy particle spectrum is .
one generally assumes that it is just the standard model particle spectrum , but there could exist light particles , completely neutral under the standard model gauge group , that interact with the standard model particles only through the exchange of new heavy particles .
this is not at all an exotic situation since it is what happened with weak interactions : neutrinos are singlets under @xmath9 and @xmath8 ; however , they can not be ignored in building an effective lagrangian that describes weak interactions . keeping in mind all these limitations
, one can very efficiently use the effective lagrangian approach to new physics to set bounds on operators that violate some of the global symmetries of the standard model : lepton number conservation , baryon number conservation or flavour symmetries @xcite .
these bounds are naturally implemented at tree level in the effective lagrangian ( although the effective operators could be generated through loops in the full theory ) . during the last years , using the precise data obtained from experiment , people have started to consider the possibility of bounding operators that do not violate any symmetry of the standard model @xcite .
this task is much more complicated because the number of operators of a given dimension is very large @xcite ( after using the equations of motion and without taking into account flavour , buchmller and wyler @xcite found 80 @xmath10 gauge invariant dimension - six operators constructed from the standard model fields ) .
especially interesting is the analysis of effective operators contributing to trilinear gauge - boson couplings , since they could have important consequences at lep2 , the ssc and lhc @xcite .
some of these operators also contribute to lep1 observables at tree level and they are strongly bounded by the lep very precise measurements .
others do not contribute at tree level to lep1 observables , but only in loops @xcite .
the next step is to use the effective lagrangian at the one - loop level and try to set bounds on all operators by using radiative corrections .
this analysis is even more complicated since the effective theory is non - renormalizable in the standard sense and care must be taken with divergences , especially when the full theory is not known . if the full theory is known , it is not difficult to find the right prescription to absorb all infinities in the effective theory .
all couplings must be renormalized , and matching to the full theory fixes all the counterterms .
if the full theory is not known , the theory still has to be renormalized and infinities absorbed in the various couplings ; however , the finite parts of the counterterms remain arbitrary and can not be determined .
the way this is done has created some controversy in the literature . the most general trilinear interaction among vector bosons
can be constructed by imposing only lorentz invariance and qed gauge invariance @xcite . in the last years
several groups have used such interactions in loops .
calculations were performed by using a momentum cut - off that was identified with the scale of new physics . then , large effects were found since , in some cases , the diagrams were quadratically or even quartically divergent .
this approach is not always correct , because the effective theory also has to be renormalized , couplings must be defined at some scale and they must satisfy some renormalization group equations .
criticisms to this treatment of effective lagrangians have already been raised by several groups @xcite ; however , the emphasis was put on different aspects .
the authors of refs .
@xcite stressed the importance of the gauge invariance of the effective lagrangian under the standard model gauge group . while , for example , in refs . @xcite the emphasis was put on the incorrect use of cut - offs in previous calculations .
the purpose of this paper is to study some of the questions that arise when the effective lagrangian approach to new physics is used by working out completely an example of possible new physics .
we consider the simplest non - trivial extension of the standard model we could write down : a standard model supplemented by a singly charged scalar singlet coupled to leptons @xcite . we construct the full model and obtain a low - energy effective field theory by integrating out , at the one - loop level , the heavy scalar singlet . by construction , since the full theory is @xmath10 gauge invariant and we integrate out a complete scalar multiplet , we automatically obtain a @xmath10 gauge invariant effective theory .
therefore , we do not discuss the role of gauge invariance in the construction of the effective theory .
we use this example to study the renormalization of the effective lagrangian and to study the matching conditions that relate the parameters of the effective lagrangian to the parameters of the full lagrangian ensuring agreement between the two theories at low energies .
the possibility of using the equations of motion before or after spontaneous symmetry breaking is illustrated as well .
the model is also phenomenologically interesting because the presence of the scalar gives rise to very interesting phenomena @xcite such as , for example , the processes @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath31 , @xmath34 , @xmath35 , @xmath31 .
there are also additional contributions to the masses of the gauge bosons , new neutral current interactions , etc .
of course , this analysis is quite far from being general , it is a very specific model that leads to a linearly realized gauge symmetry , but in spite of its simplicity it leads , already at the one - loop level , to many of the operators classified in refs .
it can also help us to understand some of the tricky points discussed in the literature . in section [ fulllagrangian ]
we will introduce the notation and write down the lagrangian of the full theory . in section [ intscalar ]
we obtain the tree - level and the one - loop contributions to the effective lagrangian of diagrams with only heavy scalars in internal lines by using functional methods .
when using the tree - level effective interactions at the one - loop level in the effective lagrangian , new effective operators appear as counterterms that must be computed by matching the full theory results with the effective lagrangian results ; we do this job in section [ matching ] .
renormalization of our effective lagrangian and the possibility of using effective lagrangians in general at the one - loop level to bound new physics are discussed in section [ renormalization ] . in section [ ssb ]
we discuss the use of the classical equations of motion to rewrite the effective lagrangian in a convenient form for phenomenological analyses .
we also discuss the impact of spontaneous symmetry breaking on this lagrangian and extract the relevant interactions in terms of the physical fields . in section [ pheno ] we shortly comment on some of the most interesting phenomenological consequences of the model by using the effective lagrangian we have obtained .
finally , in section [ conclusion ] we review what we have learned about the use of effective lagrangians with our example .
we include in appendix [ det ] the calculation of the determinant of the fluctuation operator , in appendix [ heavy - light ] the calculation of the diagrams with heavy - light lines , and finally in appendix [ fcdecay ] we calculate the amplitudes for @xmath36 in both the full and the effective theory .
[ fulllagrangian ] the complete renormalizable model we are considering is an extension of the standard model , which contains a singly charged scalar singlet in addition to the standard model particles .
this model is one of the simplest extensions one could imagine , but in spite of its simplicity , it already includes many interesting features common to any extension of the standard model containing a large mass scale compared with the fermi scale . for completeness
we list below the particle spectrum of the full model and give the corresponding transformation properties with respect to the gauge group @xmath37 + 0.5 cm [ cols= " < , < , < " , ] 0.5 cm all possible @xmath20 and generational indices are suppressed above .
the full lagrangian can be split into two parts : @xmath38 the first part , @xmath39 , represents the standard model lagrangian : @xmath40 where the covariant derivative can be written in the general form @xmath41 by @xmath42 and @xmath43 we denote the generators of the @xmath20 and the @xmath44 groups , accordingly , which are acting on the proper representations of the groups . in the case of @xmath20 doublets @xmath45 , where @xmath46 are @xmath47 pauli matrices . for @xmath44-triplets @xmath48 , where @xmath49 are @xmath50 gell - mann matrices . in the lagrangian ( [ smlagrangian ] ) the yukawa couplings , @xmath51 , are arbitrary complex matrices in flavour space .
the lagrangian @xmath52 contains the interactions of the scalar singlet : @xmath53 and the covariant derivative in eq .
( [ lsinglet ] ) has the form @xmath54 , because the @xmath55-scalar is an @xmath20 singlet with hypercharge @xmath56 . for further applications it is convenient to rewrite the @xmath55-dependent part of the lagrangian , using integration by parts for the covariant derivative , in the following form : @xmath57 the quantum numbers of the scalar singlet are such that there is no interaction between the scalar singlet @xmath55 and the quarks in the lagrangian of eq .
( [ lsinglet ] ) .
it is also important to note that its coupling to the leptons , @xmath58 , is an antisymmetric complex matrix in flavour space .
the antisymmetry of this matrix is a consequence of fermi statistics and the fact that the current @xmath59 is an @xmath20 scalar .
then , one can write @xmath60 , from which the antisymmetry of the coupling easily follows .
this property of the scalar - lepton coupling is very important since it naturally leads to a violation of the generational lepton numbers and all the interesting phenomenology related to it .
however , if every lepton carries a total lepton number of 1 , we can assign a total lepton number of 2 to the scalar @xmath55 and , then , the lagrangian ( [ lsinglet ] ) is invariant with respect to a global symmetry , which can be identified as conservation of the total lepton number . as a consequence ,
the neutrinos remain massless at all orders .
models very similar to the one defined in eq .
( [ lagrangian ] ) containing additional doublets and/or scalar singlets , in which the total lepton number is explicitly or spontaneously broken , have been used @xcite to generate small calculable neutrino masses .
[ intscalar ] if the mass @xmath61 of the scalar @xmath55 is much larger than the energy available to experiment , the only effects of this particle on the low - energy observables come through virtual corrections .
these effects can be taken into account by using an effective lagrangian of the form ( [ firsteq ] ) , containing only the `` light '' fields of the model and where the effects of the `` heavy '' particle are included in the non - renormalizable terms of dimension larger than four , which are suppressed by inverse powers of the heavy mass @xmath61 .
the effective lagrangian can be defined by integrating out the heavy scalar field @xcite : @xmath62 where @xmath63 represents functional integration over @xmath55 .
the effective field theory defined by this effective lagrangian is fully equivalent to the original theory when only green functions with light external particles are considered .
starting from eq .
( [ integraout ] ) we can calculate the one - loop level effective lagrangian by using the steepest - descent method to integrate out the heavy scalar .
as we are interested in the effects of a heavy scalar ( @xmath64 tev ) on the physics around the @xmath65 scale and below , we will systematically keep only terms of order @xmath66 through the whole calculation , neglecting all operators with higher inverse powers of the mass of the scalar singlet .
let us denote by @xmath67 the solution of the classical equation of motion for the @xmath55-field , i.e. @xmath68 using the lagrangian ( [ hlag ] ) we obtain : @xmath69 then the full action can be functionally expanded around the solution @xmath70 @xmath71 + \int d^4x d^4 x ' \eta^\dagger(x ) o(x , x ' ) \eta(x ' ) + \cdots~,\ ] ] where the fluctuation operator @xmath72 is given by the second - order term in taylor s expansion of the action @xmath73 : @xmath74 and @xmath75 in eq .
( [ expansion ] ) represents the fluctuations around the classical solution , @xmath76 . substituting eq .
( [ expansion ] ) in eq .
( [ integraout ] ) , shifting variables from @xmath55 to @xmath77 in the functional integration , and using the known formula for gaussian integration , we obtain @xmath78)\ } \det(o)^{-1 } = \exp \left\{i(s_{sm}+ s[h_0])-{\mathop{{\mbox{\rm tr}}}\left\ { \log(o ) \right\}}\right\}~.\ ] ] therefore , for the one - loop effective action we have @xmath79+i { \mathop{{\mbox{\rm tr}}}\left\ { \log(o ) \right\}}~.\ ] ] the first term in eq .
( [ seffgen ] ) is just the pure standard model contribution to the effective action . using the equation of motion , eq .
( [ eqofmotion ] ) , the second term , @xmath80 $ ] , can be formally written as @xmath81 \approx -\int d^4x { \overline{\widetilde{\ell}\ , } } ( x ) f \ell(x ) \frac{1}{(-d^2-m^2-\beta \varphi^\dagger(x ) \varphi(x ) ) } \overline{\ell}(x ) f^\dagger \widetilde{\ell}(x)\ , \ ] ] where terms of order @xmath82 have been neglected .
the lagrangian ( [ lnonlocal ] ) is highly non - local . to obtain a local version of it
we could further expand the operator@xmath83 as a power series in @xmath84 : @xmath85 keeping the first term we obtain the only tree - level contribution , at order @xmath84 , of the scalar to the effective lagrangian : @xmath86= \int d^4 x { \cal l}^{(0)}(x)\ ] ] with @xmath87 where the summation over repeated flavour indices ( @xmath88 ) is assumed .
it could also be obtained by computing the tree - level diagram of fig .
1.a in the limit @xmath89 ( where @xmath90 is the momentum of the scalar ) .
the result in eq .
( [ lagr0 ] ) is not complete if we are going to use this lagrangian for calculations at the loop level @xcite .
the reason is that in order to obtain a local lagrangian we had to use the expansion ( [ appsolution ] ) .
in doing this we assumed that the derivative in eq .
( [ appsolution ] ) , or equivalently the momentum of the scalar , is negligible in comparison with the mass of the scalar .
but this is not true if the @xmath55-scalar contributes inside loops , since in this case its momentum runs up to infinity .
it has been shown @xcite that the procedure just outlined can be justified as long as one includes in the effective lagrangian a set of local operators that compensate for the terms missed by using expansion ( [ appsolution ] ) .
we are going to compute these operators in the next section .
the last piece of the effective action ( [ seffgen ] ) is defined in terms of the fluctuation operator ( [ fluct ] ) and takes into account all one - loop effects with only the heavy particle in loops . in our approximation , keeping only terms @xmath91 , the fluctuation operator ( [ fluct ] ) can be easily calculated and leads to the following effective action @xmath92 we moved to appendix [ det ] the detailed evaluation of the determinant of a generic operator of this form as an expansion in @xmath84 . using the general result of appendix [ det ] , eq .
( [ resdet ] ) , for our particular case , @xmath93 , @xmath94 , that is @xmath95 and @xmath96 , and taking into account that @xmath97 , @xmath98 , we write this contribution to the effective lagrangian as a sum of two parts : @xmath99 the first part , @xmath100 , contains only operators with dimension not larger than four .
it is given by the following expression : @xmath101 all the coefficients in the above lagrangian are ultraviolet - divergent . here and in the rest of the paper we use dimensional regularization , where these divergences appear as simple poles , @xmath102 , ( @xmath103 ) in the function @xmath104 defined in eq .
( [ deltae ] ) of appendix [ det ] .
as expected , all terms in eq .
( [ lrenorm ] ) are already present in the standard model lagrangian .
therefore , they can be absorbed in a redefinition of the parameters of the standard model .
in the second part of eq .
( [ lfluctop ] ) , @xmath105 , we included all dimension - six operators : @xmath106 the coefficients of these operators are free of divergences and are suppressed by @xmath84 .
note that the last term in eq .
( [ resdet ] ) , which is trilinear in the field strength , is zero in the abelian case .
it is clear that one could also obtain the interactions in eq .
( [ lrenorm ] ) and eq .
( [ lagr1 ] ) by using ordinary feynman rules . in fig .
2 we give the diagrams that give rise to this part of the effective lagrangian .
functional methods provide , however , a much more elegant and compact method to calculate them . as mentioned before
, all the terms in @xmath100 can be absorbed in a redefinition of the standard model parameters .
the first term in eq .
( [ lrenorm ] ) is only a renormalization of the vacuum energy .
we can neglect it , or , if we prefer , it can be absorbed in the constant term of the higgs potential of the standard model .
the second term renormalizes the quadratic coupling in the higgs potential : @xmath107 the third one is a renormalization of the quartic coupling of the standard higgs : @xmath108 finally , the fourth term in eq .
( [ lrenorm ] ) , which is an additional contribution to the kinetic term for the @xmath109-field , can be removed by wave - function renormalization : @xmath110 in order to keep the canonical form of the covariant derivative , we have to renormalize also the @xmath111 coupling @xmath112 keeping the product @xmath113 invariant : @xmath114 the above equations ( [ barmatchm ] ) , ( [ barmatchlambda ] ) and ( [ barmatchg ] ) relate the bare parameters of the full lagrangian @xmath115 , @xmath116 , and @xmath112 to the corresponding bare parameters @xmath117 , @xmath118 and @xmath119 of the effective lagrangian .
[ matching ] as commented in the previous section , it is necessary to expand the operator of eq .
( [ appsolution ] ) in powers of @xmath84 to have a local effective lagrangian , but this expansion is only valid at the classical level , since the limit of @xmath120 and the functional integration over the light fields do not commute .
this is just a consequence of the fact that the momentum @xmath121 of the scalar inside loop diagrams runs up to infinity and then an expansion in @xmath122 is not appropriate . in spite of this problem
it is still possible to obtain a local effective lagrangian as an expansion in @xmath84 @xcite .
it is necessary , however , to include some additional operators that are not expected from the nave expansion in eq .
( [ appsolution ] ) .
these operators appear as a result of quantum corrections , hence , they are suppressed by additional couplings and @xmath123 factors .
the practical way to obtain these terms of the lagrangian is to consider the most general linear combination of all the operators with a given dimension , allowed by symmetry , and then to extract the coefficients of this combination by matching the effective lagrangian calculation to the full theory calculation .
most of the operators obtained should be included in any case as counterterms to cancel the divergences generated when the tree - level lagrangian of eq .
( [ lagr0 ] ) is used inside loops .
but it is important to note that the finite parts of the coefficients of those operators can only be fixed by computing the various green functions with both the full and the effective theories and matching the results for small energies compared with the heavy mass .
some other operators , however , appear with finite coefficients .
they are just a consequence of the matching procedure and can not be obtained from the divergences that appear in the effective theory . in the diagrammatic language
all these new operators are generated by feynman diagrams , with both the heavy scalar and lepton lines in the loops .
we will compute in the full theory the amplitudes corresponding to these diagrams and will compare them with the equivalent amplitudes obtained by using the tree - level effective lagrangian , eq .
( [ lagr0 ] ) , in one - loop diagrams .
the difference will give us the necessary counterterms .
actually , the effective lagrangian calculation can be avoided by splitting the scalar propagator in two parts : @xmath124 if we use this form when calculating diagrams with only one scalar propagator , the first part gives exactly what one would obtain by using the effective tree - level four - fermion interaction , eq .
( [ lagr0 ] ) , while the second part gives just the counterterm we should add to the effective lagrangian .
we would also like to note that by splitting the propagator in these two pieces we have increased the degree of ultraviolet divergence in each of the two terms , with respect to the original diagram . on the other hand ,
the second part contains an additional factor @xmath125 in the numerator , which reduces the infared degree of divergence of this contribution .
therefore , any possible small momentum singularity is transmitted to the low - energy effective lagrangian , as it should be , since infared singularities have nothing to do with the high - energy behaviour of the theory . a technical issue about the calculation is the selection of diagrams one should compute to reconstruct the full effective lagrangian .
the effective lagrangian must be a linear combination of gauge - invariant operators .
each of the operators gives rise to a variety of physical processes . obviously , to obtain the coefficients of these operators
it is not necessary to compute all these processes , since many of them are related by gauge invariance .
our strategy consists in computing all one - particle - irreducible diagrams with the minimal number of external particles .
we keep track of all external momenta ( of course only to the order @xmath122 ) without using any equation of motion for the external particles . after that
, we write an effective lagrangian that reproduces those amplitudes and when there is no ambiguity , we reconstruct gauge invariance by promoting any derivative to a covariant derivative . sometimes , however , there is some ambiguity in the promotion of a derivative to a covariant one .
then , we compute also the diagrams with one additional external gauge boson in order to disentangle that ambiguity and we check that the gauge - invariant effective lagrangian correctly describes all the amplitudes . finally , as an additional check , we compute some diagrams with three external gauge bosons by using both the full and the effective lagrangians .
we do it for the special case of zero external momenta . ] .
since the full calculation is tedious , we give the details in appendix [ heavy - light ] , presenting here only the main results and an explanation of the procedure .
we give in fig .
3 the diagrams contributing to the lepton wave function renormalization in the full and the effective theories . in the effective theory
the wave function renormalization is zero because it is a massless tadpole - like diagram ( fig .
however , in the full theory there is a non - zero contribution that must be included as an additional effective operator . from the results of appendix [ heavy - light ] we find that , for the first term in eq .
( [ selfe ] ) , the additional operator we should include is @xmath126 where we have defined the matrix @xmath19 as @xmath127 note that even though in appendix [ heavy - light ] we have only calculated the charged scalar contribution to the self - energy of the doublet , gauge invariance requires that the partial derivative should be substituted by a covariant derivative . in this case the promotion from the derivative to the covariant derivative can be done without ambiguity .
this leads to eq .
( [ selfenergy ] ) .
it implies that there should be an additional contribution to the coupling of the gauge bosons to the lepton doublet given by the coefficient in eq .
( [ selfenergy ] ) .
we will see in the next subsection that , indeed , this contribution appears . the second term in eq .
( [ selfe ] ) is proportional to @xmath128 , which requires an effective lagrangian of the form @xmath129 .
however , the promotion of this term to covariant derivatives is ambiguous : should we use @xmath130 , @xmath131 or @xmath132 ?
the only way to resolve this ambiguity , as we will see below , is to perform a full calculation with one external gauge boson .
the operator in eq .
( [ selfenergy ] ) contributes to the kinetic term of the doublets . to write it again in a canonical form
, we should redefine the lepton doublet as follows : @xmath133 this redefinition only affects the tree - level four - fermion effective lagrangian and the standard model yukawa coupling .
this is because the wave function renormalization appears at the one - loop level , and , as a consequence , its effect on all the other ( one - loop ) operators is a two - loop effect .
it can be taken into account by redefining the standard model yukawa couplings and the couplings @xmath134 that appear in eq .
( [ lagr0 ] ) as follows @xmath135 as we see this does not change the flavour structure of the couplings , since the @xmath136 is still antisymmetric in flavour indices and does not give any new interesting process .
the same thing could be said about the standard model yukawa couplings @xmath137 . in fig .
4 we present the contributing diagrams to the vertex of the @xmath109 and @xmath138 gauge bosons with leptons , with the heavy scalar running in the loop ( figs . 4.a and 4.b ) .
note that for the @xmath138 there is no diagram fig . 4.b , because the scalar is an @xmath20 singlet .
we also give the corresponding diagrams in the effective theory ( fig .
here it is interesting to note that the amplitude corresponding to the diagrams of figs .
4.a and 4.b in the full theory is finite , when the diagrams with external wave function renormalization are included .
this can be understood because the full theory is renormalizable : since all these diagrams lead to a flavour - changing vertex and as our original lagrangian does not contain such a coupling , there is no counterterm available . in the effective theory language , this cancellation is just a consequence of gauge invariance : the divergent contributions to the vertex in the full theory can be taken into account by the gauge - invariant dimension - four operator of eq .
( [ selfenergy ] ) , which can be removed by a wave function renormalization of the lepton doublet .
we give in appendix [ heavy - light ] the resulting calculation , eqs .
( [ ampb ] ) and ( [ ampw ] ) , of the vertex diagrams of the @xmath109 and @xmath138 after subtraction of the effective lagrangian contribution ( diagram 4.c ) .
we cast this result in a form that can be easily identified with some effective operators .
the first term can already be obtained , as promised , from eq .
( [ selfenergy ] ) . the rest of the terms can be obtained ( for both the @xmath109 and the @xmath138 fields ) from the following operators : @xmath139 in these equations , @xmath140 is the appropriate covariant derivative for each of the fields .
it is easy to see that the pure derivative part of the first term eq .
( [ selfenergy2 ] ) , reproduces perfectly the second term in eq .
( [ selfe ] ) and that the ambiguity mentioned in the previous subsection has been solved completely with this calculation . to check this result further ,
we have computed the diagram with three @xmath141 gauge bosons for zero external momenta and compared the result with that obtained from eqs .
( [ selfenergy2])([penguin2 ] ) for constant fields .
the physical content of these operators is quite obscure .
then , we found it convenient to use the operator identity : @xmath142 to rewrite eqs .
( [ selfenergy2])([penguin2 ] ) in the following form : @xmath143 as we will see later , after using the equations of motion , the physical content of these operators becomes more transparent .
the first term can be removed in favour of yukawa - type couplings and does not give any interesting process .
the second and the third terms will give rise to processes such as @xmath144 and @xmath34 , and finally the last terms , with the appropriate combination of @xmath145 and @xmath109 , will give rise to @xmath32 .
there are also some operators that involve two scalar higgses , two lepton doublets , and one covariant derivative .
they are generated by the diagrams of fig .
5 and they do not contribute to any interesting process . for completeness
we give the result here . from the diagram of fig .
5.b we get the amplitude eq .
( [ amp5a ] ) , which can be obtained by using the operator @xmath146 from the diagram of fig . 5.a , and after subtraction of the effective lagrangian contribution , fig . 5.c , we get the amplitude ( [ amp5b ] ) , which can be obtained from the operators @xmath147 where @xmath148 we used an @xmath20 fierz transformation to write the operators in this form , being the original operators expressed in terms of @xmath149 .
clearly the effects of operators ( [ op5b ] ) are suppressed by at least two powers of the lepton masses . in this section
we shall discuss possible new four - fermion interactions that could arise at the one - loop level .
these interactions can come from three different kinds of diagrams : \i ) diagrams with gauge boson corrections to the vertex of the charged scalar with the lepton doublet .
since these only contain light particles in the loop , they are fully taken into account by gauge boson corrections to the effective four - lepton interaction ( [ lagr0 ] ) .
in addition they do not change the flavour structure of the coupling .
\ii ) box diagrams with a charged scalar and a gauge boson running in the loop ( fig .
it is easy to see that diagrams with a @xmath138 exchange just cancel because of the antisymmetry of the couplings @xmath134 .
diagrams with exchange of @xmath109 renormalize the tree - level four - lepton interaction without changing its flavour structure .
part of their contributions are taken into account by the effective theory diagrams of fig .
but , obviously , these diagrams do not modify the flavour structure of the tree - level effective four - fermion interaction ( [ lagr0 ] ) and , then , they can not give rise to processes such as @xmath150 at least at order @xmath84 .
\iii ) box diagrams with two charged scalars in the loop ( fig .
diagrams of this type are finite and in the sum they give rise to the following operator : @xmath151 it can be shown that this operator does not contribute to processes with two identical fermions , such as @xmath152 . indeed , taking only into account , for example , the down component of the @xmath20 lepton doublets , the interaction among four charged leptons induced by eq .
( [ boxb ] ) can be written as @xmath153 where , to obtain the right - hand side , we have used a fierz transformation .
we immediately see that this operator does not contribute to processes such as @xmath152 .
however it does contribute to other interesting processes such as @xmath154 . clearly
only the contributions of type ( iii ) are interesting and only these will be included in our effective lagrangian .
before we summarize the results of section 4 , we note that all the results obtained in section [ matching ] have been derived from the effective tree - level lagrangian given in eq .
( [ lagr0 ] ) .
one can easily see that this lagrangian can also be written ( after a combined fierz transformation in @xmath20 and dirac space ) in the more familiar form @xmath155 but this is true only in four dimensions : the fierz transformations we used above can not be done in @xmath156 dimensions .
should we start with this new lagrangian , the finite parts of the counterterms would be different when using dimensional regularization , but this formulation would be completely equivalent to the original one .
we decided to keep the original form of the lagrangian so as to keep the notation as close as possible to the full lagrangian formulation .
using the results of sections 3 and 4 , the complete one - loop effective lagrangian is @xmath157 here @xmath39 is the standard model lagrangian , eq .
( [ smlagrangian ] ) , @xmath158 as given in eq .
( [ lagr0 ] ) is the tree - level @xmath84 term of the effective lagrangian , @xmath105 is given in eq .
( [ lagr1 ] ) and contains all @xmath84 one - loop contributions of diagrams with only singlet scalars in the loop and finally @xmath159 , which is the sum of the terms in eqs .
( [ selfenergy2a])([op5b ] ) and ( [ boxb ] ) , gives all the contributions coming from matching the full theory in diagrams with heavy - light particles in the loops : @xmath160 all the couplings in this lagrangian should be understood as bare barred effective lagrangian couplings related to the full theory bare couplings through eqs .
( [ barmatchm ] ) , ( [ barmatchlambda ] ) , ( [ barmatchg ] ) , ( [ redefy ] ) and ( [ redeff ] ) . to simplify the notation we have suppressed everywhere generational indices , then @xmath19 and @xmath161 are the matrices defined in eqs .
( [ deff ] ) and ( [ defhatf ] ) .
[ renormalization ] equations ( [ barmatchm ] ) , ( [ barmatchlambda ] ) , ( [ barmatchg ] ) , ( [ redefy ] ) and ( [ redeff ] ) relate the bare couplings and masses of the effective theory with those of the full theory .
it is , however , more interesting to have the equivalent relations for renormalized couplings .
if an @xmath162 scheme is employed to renormalize both the full and the effective theories , we can very easily obtain the matching equations for the renormalized couplings .
let us take , for example , the gauge coupling @xmath112 .
we denote the @xmath162 renormalized quantities with the same symbol as the bare quantities , but adding an additional dependence on the renormalization scale @xmath163 .
all effective theory quantities will be distinguished by a bar .
the standard relationship between bare and renormalized couplings in @xmath162 schemes is ( @xmath164 and @xmath165 ) : @xmath166 where @xmath167 and @xmath168 are the lowest - order coefficients of the @xmath169-functions for the coupling constants in the full and the effective theories , respectively . substituting these equations in eq .
( [ barmatchg ] ) and equating finite terms @xcite we obtain the desired matching condition for renormalized couplings : @xmath170 note that this equation can be obtained by just dropping the @xmath171 contained in @xmath172 in eq .
( [ barmatchg ] ) .
this is not surprising since the divergent term in eq .
( [ barmatchg ] ) gives just the charged scalar contribution to the beta function of @xmath112 in the full theory .
similar arguments give for the rest of the couplings and masses the following result @xmath173 @xmath174 @xmath175 and a matching equation for the four - lepton coupling in the effective theory expressed in terms of the full theory couplings and masses .
it can be obtained by using an effective @xmath176 defined as follows : @xmath177 in this equation there could also be some contributions from box diagrams , which we have not computed .
these matching conditions are , in principle , valid for an arbitrary value of the renormalization scale @xmath163 .
however , it is clear that in order to avoid large logarithms they should be evaluated at some scale around the charged scalar mass and then , using the standard model renormalization group , run all the couplings to obtain their values at lower scales . equation ( [ matchm ] ) is very interesting , since it manifests in all its crudeness the so - called naturalness problem of the standard model ; @xmath178 is the mass parameter that appears in the higgs potential part of the effective lagrangian , and it has to be of the order of the electroweak scale .
however , it is clear that if we take @xmath179 very large , we should also take @xmath180 large in order to have @xmath178 small enough .
but even if we do so at some scale @xmath163 , it will be very difficult to keep @xmath178 small at any other scale .
this represents a serious fine - tuning problem , which appears when the standard model is embedded in another model containing mass scales much larger than the fermi scale .
it is important to note that by using dimensional regularization the problem appears only in the matching conditions .
if a cut - off regularization scheme is used , the naturalness problem can be related to the appearance of quadratic divergences . however , this relation is not direct .
quadratic divergences can appear at the intermediate stages , but they should be absorbed in a renormalization of the mass parameters of both the full and the effective theories , since physical quantities must be cut - off - independent .
after renormalization , the naturalness problem should also appear only as a fine - tuning problem in the matching conditions , as in eq .
( [ matchm ] ) .
as the main physical consequences can be obtained equally in dimensional regularization , we do not find any particular advantage by working with a cut - off regularization scheme . if an @xmath162 scheme is used to renormalize also the higher dimension operators , we could write the effective lagrangian in terms of a set of running couplings @xmath181 for each of the operators @xmath182 where the @xmath181 are obtained from the bare coefficients in eq .
( [ totalefflagr ] ) just by dropping the @xmath183 terms in @xmath172 ; that is just by substituting @xmath172 by @xmath184 .
then all couplings should be understood as renormalized effective lagrangian couplings ( barred couplings ) related to the full theory couplings through eqs .
( [ matchg])([matchf ] ) .
to simplify matters we can consider only processes that violate muon - lepton number in one unit and electron - lepton number in minus one unit ( the total lepton number must be conserved ) .
for example , we can take the two operators @xmath185 the factor @xmath186 in the penguin operator has been included for convenience . from eqs .
( [ lagr0 ] ) and ( [ penguin1 ] ) , we obtain expressions for the corresponding running couplings : @xmath187 where the dots represent additional one - loop contributions , which , as we will see immediately , are not important at the level we are working .
note that @xmath188 because of the antisymmetry of the couplings in flavour indices .
equations ( [ cesp])([ces ] ) can be cast into the following form @xmath189 with @xmath190 and the couplings at the scale @xmath191 are given by @xmath192 since @xmath193 appears only at the one - loop level in the full model , the last term in eq .
( [ c1 ] ) is a two - loop effect .
this is the reason why we have not considered it in eq .
( [ cesp ] ) . on the other hand , if @xmath163 is not very different from @xmath61 we have @xmath194 and @xmath195 ; then , the diagonal elements of the anomalous dimension matrix do not need to be computed .
equations ( [ c1 ] ) and ( [ c2 ] ) are an approximate solution of the general renormalization group equation that describes mixing among the operators @xmath196 and @xmath197 @xmath198 which is valid only in the case that @xmath199 . if we pretend to use this solution for a wider range of @xmath163 s , the full @xmath200 matrix should be computed and the integration of eq .
( [ renopmix ] ) should be done by taking also into account the running of the gauge coupling constants . however , for the standard model values of the gauge coupling constants the linear approximation ( @xmath201 ) works very well for a wide range of scales @xmath202 .
obviously , in this approximation and taking into account that only the operator in eq .
( [ lagr0 ] ) is generated at tree level it is clear that in our model it is enough to consider only mixing of the effective operators generated at the one - loop level with the tree - level operator because other mixings would represent two - loop effects .
then , at the order we are working , we have always @xmath203 operator mixing .
this simple example shows us clearly what can and what can not be obtained by using the effective lagrangian . from the effective theory
we could calculate the anomalous dimension matrix that controls the mixing among the different operators in the effective lagrangian since the logarithmic terms are the same in the full and in the effective theories , however the boundary conditions for the renormalization group equation ( [ renopmix ] ) can only be obtained after the matching procedure
. if we do not know the full theory a set of boundary conditions can only be obtained from experiment . in this case
we can still compute the anomalous dimensions in the effective theory , write down eq .
( [ renopmix ] ) , and solve it for arbitrary boundary conditions at the scale @xmath204 .
then the solution , within our previous approximations , is @xmath205 now we should take the full , complete basis of operators that mix at the one - loop level , since if the boundary conditions are not known we can not neglect _ a priori _ any of the mixings . to keep simplicity , however , we will discuss only two - operator mixing .
the interesting point about eqs .
( [ ces3 ] ) is that , if we know all the effective couplings at some scale @xmath204 , we can obtain them at some other scale ( not necessarily the scale responsible for the new physics ) . equations ( [ ces3 ] ) clearly tell us that to predict the values of the couplings at any scale we need two ( in the case of mixing of two operators ) boundary conditions .
these conditions could be obtained from experiment .
for example , if we could bound the couplings of the two operators at lep1 we could straightforwardly predict these couplings at lep2 energies .
however , if in our example we only know one of the couplings at lep1 , no matter how many loops we use to calculate the anomalous dimensions , we will not be able to bound the other operator at the same or any other scale .
only if the full theory is known can we relate the couplings of different operators as they are expressed in terms of the few parameters of the full theory .
then , bounds on one operator can be related to bounds on other operators . in our example
it is clear from eq .
( [ boundary ] ) that @xmath206 or , for instance at the @xmath65 scale , by using eqs .
( [ cesp])([ces ] ) we have @xmath207 then it is clear that a bound on @xmath208 implies a bound on @xmath209 , and viceversa , and it also implies a bound on @xmath210 and @xmath211 at any other scale @xmath163 .
but it is important to realize that this is only possible because our full theory provides us with additional relationships among couplings . of course
, one can always impose additional assumptions to obtain an estimate of the bounds on different operators .
for example , if one assumes that there are no unnatural cancellations among the couplings of different operators one can bound each of them independently from the others . in our example
this would be implemented by putting , for instance , @xmath212 and @xmath213 , then use eqs .
( [ c1 ] ) and ( [ c2 ] ) with @xmath214 to obtain @xmath208 and @xmath209 as a function of @xmath215 .
clearly with this assumptions a bound on @xmath209 implies a bound on @xmath215 and therefore also a bound on @xmath208 . and similarly for the other coupling .
the estimates so obtained are interesting and can be useful .
however , we feel that they can never substitute more elaborate bounds based on a complete set of experimental data .
[ ssb ] from all previous sections it is clear that our effective lagrangian reproduces the results of the full theory at the one - loop level as long as energies smaller than the @xmath55 mass are involved .
however , until now we have considered the full unbroken theory , but the standard model and the extension of it we are considering now undergo spontaneous symmetry breaking ( ssb ) and this has to be taken into account in the analysis .
another point that has been the origin of controversy is the possibility of using the standard model equations of motion to reduce the number of operators in the effective lagrangian or to make its physical content more evident .
it is well known that the equations of motion can not be used navely in a lagrangian if it is going to be used to calculate diagrams with particles off - the - mass - shell .
in fact , the possibility of using the equations of motion to trade derivative couplings in favour of pseudo - scalar couplings has led to many wrong calculations in pion physics and in other models containing goldstone bosons .
however , the equations of motion can still be used under certain circumstances .
first of all , if the lagrangian is going to be used to calculate tree - level amplitudes with all particles on - the - mass - shell , they can be used without any problem .
moreover it can be shown @xcite that the use of the equations of motion is equivalent to a redefinition ( in general with a non - linear transformation ) of the fields of the theory . if the starting theory is renormalizable , normally these non - linear transformations lead to lagrangians that contain operators of higher dimension and which are not explicitly renormalizable . however , if the starting theory is described by an effective lagrangian , with all kinds of non - renormalizable interactions , the effect of these transformations is equivalent to the use of the equations of motion plus a modification of the higher - order terms in the effective lagrangian .
this is not a problem since all these terms are already present in the effective lagrangian .
then , it only amounts to a re - ordering of the effective lagrangian @xcite .
there are some subtleties related to the jacobian of the transformation and renormalization , but in general it can be shown that they do not represent a problem @xcite .
however , one has to be careful when studying processes that are sensitive to operators of different dimensions . in our case
there is no problem since the only operators that could be eliminated with the equations of motion appear at the one - loop level ; then , they are going to be used uniquely as tree - level insertions .
we could use the equations of motion before or after ssb , or not use them at all , but the final result should be independent from this . intermediate steps , however , can look quite different .
we found it convenient to use the equations of motion of the standard model before ssb , because they are simpler and because it makes it easier to see which new physics is generated by the effective lagrangian .
we are going to use them to substitute the derivatives of field strengths in favour of fermionic currents and new gauge boson interactions involving the standard higgs scalar .
similarly , we will substitute covariant derivatives of the lepton doublet in favour of the right - handed leptons and the standard higgs . later on we will see , in a particular example , that physical observables are the same as the results one would have obtained if proceeding in a different way . from the standard model lagrangian , eq .
( [ smlagrangian ] ) , one can easily obtain the following equations of motion for the electroweak gauge bosons @xmath216 where @xmath217 are the hypercharge and @xmath20 fermionic currents)([currentw ] ) differ from eqs .
( 2.13 ) and ( 2.14 ) in ref .
@xcite , where some terms are missing . ] .
we are also going to use the standard model equation of motion for the leptonic doublet @xmath218 we will use eqs .
( [ eqmotb])([eqmotw ] ) and ( [ eqmotdoublet ] ) in ( [ totalefflagr ] ) .
every time we use the equation of motion for the leptonic doublet we obtain an additional leptonic yukawa coupling constant @xmath219 .
since these couplings are small , we are going to neglect terms that contain two or more of those yukawa couplings . after using ( [ eqmotb ] ) , ( [ eqmotw ] ) and ( [ eqmotdoublet ] ) in @xmath220 , the resulting lagrangian is @xmath221 the terms from eq .
( [ totalefflagr ] ) containing @xmath131 and @xmath161 have been neglected as they are suppressed by two or more leptonic yukawa couplings .
again all couplings must be understood as bare effective lagrangian couplings ( barred couplings ) .
we can write eqs .
( [ afterem1])([afterem7 ] ) in terms of @xmath162 renormalized couplings by just dropping the divergent contributions ( @xmath222 ) . it is clear that after using the equations of motion one can easily see the plethora of interesting processes that a single term in eq .
( [ totalefflagr ] ) can generate .
we will first study how our effective lagrangian can modify the standard model mechanism for ssb and what its effects are on the spectrum of physical gauge bosons . the first term in ( [ afterem1 ] )
modifies the standard model higgs potential and produces a shift in the vacuum expectation value ( vev ) .
since this is a one - loop effect one should also use the full one - loop standard model effective potential to analyse this shift .
this gives a global redefinition of the vev and has only consequences in the higgs sector of the theory : the ratio of the higgs to the @xmath223 mass and the higgs couplings are changed , but there are no other effects .
as we are not interested in higgs physics here we are not going to compute this shift .
the higgs dependent terms of the effective lagrangian can have interesting effects when the higgs develops a vev : @xmath224 in what follows we will substitute the higgs field by its vev and will neglect all the terms containing physical higgses . the second term in eq .
( [ afterem1 ] ) , after ssb , produces a wave function renormalization of the higgs scalar . again
, since we are not interested in higgs interactions here we are going to neglect it .
the third term clearly gives a wave function renormalization of the @xmath109 gauge boson .
then , one has to diagonalize simultaneously the kinetic term and the mass terms for the gauge bosons .
after ssb we get the following kinetic term for the @xmath109 gauge boson @xmath225 where @xmath226 the @xmath223 gauge boson kinetic term and the mass terms remain unchanged . then we can recover the canonical kinetic term by redefining the @xmath109 field as follows @xmath227 this will change the mass matrix of the gauge bosons .
however as the @xmath109 field comes always with @xmath112 we can recover the standard model form by redefining @xmath112 as well @xmath228 it is not difficult to see that after ssb the last term in eq .
( [ afterem2 ] ) gives @xmath229 where we have already included the redefinition of @xmath112 from eq .
( [ hatg ] ) .
this term is an additional contribution to the neutral gauge bosons mass term .
adding it to the standard model contribution and diagonalizing the full gauge boson mass matrix , we find that the @xmath145 and @xmath109 gauge bosons can be expressed in terms of the physical photon @xmath230,and the @xmath0-boson @xmath231 as follows : @xmath232 where @xmath233 and @xmath234 is the tree - level weak mixing angle in the pure standard model .
similarly the relation between the electric charge and the gauge coupling is modified in comparison with the standard model one : @xmath235 with the same notation the physical masses are @xmath236 where @xmath237 comes from eq .
( [ ncoperators ] ) . from these equations
it is clear that only the @xmath238 correction , which produces a relative shift form the standard relation between the masses of the @xmath0 and the @xmath223 could in principle be observed ; however , it is very small .
all the other shifts related to @xmath239 can be absorbed in the definition of the coupling @xmath240 or , equivalently , in the weak mixing angle @xmath241 and then are not observable .
now we can go back to eqs .
( [ afterem1])([afterem7 ] ) and write them in terms of the physical gauge bosons .
the result we have found ( neglecting all higgs interactions ) is : @xmath242 here @xmath243 is the charged lepton mass matrix and @xmath244 and @xmath245 are the field strengths of the photon and the @xmath0 gauge boson , respectively . @xmath246 and @xmath247 are the components of the left - handed lepton doublet and @xmath248 are the right - handed parts of the charged leptons .
we have also defined the following neutral currents : @xmath249 where @xmath250 and @xmath251 is the usual charged current .
couplings with more than one gauge boson have not been included , and the term ( [ afterem6 ] ) has been removed because after ssb it can be absorbed in a renormalization of the charged lepton mass matrix . as we have seen , by using the equations of motion we have obtained a lagrangian that displays its physical content in a very transparent way .
however , as stressed before , the use of the equations of motion is not necessary .
we could have started with the unbroken lagrangian , eqs .
( [ totalefflagr ] ) and ( [ lagr1 ] ) , and then let the higgs field acquire a vev without using the equations of motion at all .
after diagonalization of the gauge boson mass matrices we would have obtained a quite different effective lagrangian from that in eqs .
( [ current - current])([extrabox ] ) . however , both lagrangians give the same physics ( at least at the level of precision we are working here ) .
for example , the penguin operators in @xmath252 ( that is the first two operators in the lagrangian of eq .
( [ totalefflagr ] ) ) when written in terms of the physical fields , give the following interactions between @xmath0-bosons and leptons : @xmath253 clearly , from this interaction and for on - the - mass - shell particles , we get the same amplitude for the @xmath0 decay into a pair of different leptons as that obtained from eqs .
( [ fczncoupling ] ) and ( [ fczecoupling ] ) . from eq .
( [ zpenguin ] ) we see that @xmath0-exchange processes at low - energy , @xmath254 , are suppressed by @xmath255 .
we get exactly the same answer from the compensation of the contributions from the two terms ( the @xmath256 and the @xmath257 current terms ) in eq .
( [ fczncoupling ] ) .
[ pheno ] the terms in @xmath252 give many interesting effects , apart from the shifts in the gauge - boson masses , which have already been commented on .
for example , eq . ( [ current - current ] )
gives extra contributions to flavour - conserving neutral - current processes .
equation ( [ emfccoupling ] ) gives rise to four - fermion processes with non - conservation of generational lepton numbers ; for example they give rise to @xmath258 and similar processes .
to compute these one has to take into account also the contributions coming from one - loop diagrams , with one insertion of the tree - level four - lepton interaction .
the couplings in eq .
( [ fczecoupling ] ) allow the decay of the @xmath0 into different charged leptons .
equation ( [ muegamma ] ) leads to @xmath32 and similar processes .
the amplitude for this process can easily be obtained from eq .
( [ muegamma ] ) . taking into account
that , without loss of generality , @xmath259 can be taken diagonal with diagonal elements @xmath260 it is given by @xmath261 @xmath262 and @xmath263 are , respectively , the left - handed and right - handed chirality operators . the amplitude eq .
( [ ampmuegamma ] ) is in complete agreement with the results obtained in refs .
@xcite .
we are not going to study in detail the phenomenology of this model here .
an analysis of this based on our effective lagrangian , will be presented elsewhere . however , to give a flavour of the applicability of our effective theory , and to show how one can work with it , we will comment on the calculation of the decay @xmath264 with both the full and the effective theories . in the effective theory we have two amplitudes , one due to the loop diagram of fig .
4.c and the other coming from eq .
( [ fczecoupling ] ) .
as expected , even though each of them contains a divergent contribution @xmath172 , they give a finite answer when summed .
the total amplitude is given in appendix [ fcdecay ] , eq .
( [ efffcamp ] ) .
we also have performed a complete calculation of the amplitude of this process in the full theory . after taking into account the diagrams with self - energy insertions to the external fermion legs ,
the final answer is finite .
the amplitude is given in eq .
( [ fullamp ] ) in terms of the functions @xmath265 and @xmath266 , as explained in appendix [ fcdecay ] . in the limit of @xmath267 ( that is , @xmath268 ) ,
the results obtained in the full and in the effective theories are identical , as they should be .
in fact all matching conditions are designed with just this in view . from the full model amplitude of eq .
( [ fullamp ] ) , we obtain the following branching ratio of the flavour - violating decay width ( @xmath269 ) relative to the standard model flavour - conserving one , @xmath270 the branching ratio in the effective lagrangian is given by the same expression , but with @xmath265 and @xmath266 given in eq .
( [ limfes ] ) . in fig .
7 , we present this branching ratio in both the full and the effective theories as a function of @xmath61
. charged scalar yukawa couplings , @xmath271 , are taken to be equal to 1 . from the figure it is clear that for masses of the scalar @xmath55 larger than the mass of the @xmath0 boson the effective theory gives a good approximation .
[ conclusion ] in this paper we have studied some questions related to the construction and the use of effective lagrangians , by considering an extension of the standard model that includes a heavy charged scalar coupled to the leptonic doublet . starting from the full renormalizable model , we have built a low - energy effective field theory by integrating out the heavy scalar .
this was done at the one - loop level and keeping only operators of dimension six or less .
functional methods were used to obtain , in a compact and gauge - invariant form , all operators generated by tree - level and one - loop diagrams containing only heavy scalars in the internal lines .
the result includes only diagrams which are one - particle - irreducible with respect to the light particles . in appendix [ det ]
we give a detailed explanation of this calculation .
this is not the complete answer for the effective lagrangian , since there are many one - loop diagrams in the full theory , with heavy and light particles in the loop that are not completely taken into account by one - loop diagrams involving tree - level effective couplings . to obtain these additional contributions
we have calculated several green functions in both the full and the effective theories and required that the results of both calculations match for small momenta compared with the heavy scalar mass .
the result can be expressed as a linear combination of gauge - invariant operators , which must be included in the effective lagrangian . adding both contributions , from the operators corresponding to the diagrams with only heavy lines and from those with heavy and light internal particles , we have obtained the complete bare one - loop effective lagrangian , including operators up to dimension six .
it contains several operators with infinite coefficients ( dimensional continuation was employed to regularize all divergent integrals , then uv divergences appear as poles in @xmath272 ) . by using an @xmath162 scheme to renormalize both the full and the effective theories
, we obtained the matching conditions for the running couplings , which express the renormalized couplings of the effective theory in terms of the renormalized couplings of the full model . to avoid large logarithms ,
these matching conditions should be evaluated at a scale around the heavy scalar mass and then the renormalization group used to bring the couplings to any other lower scale .
we used this simple example to discuss the behaviour of the couplings of a generic effective lagrangian .
in general all the operators in the effective lagrangian with the same quantum numbers mix under the renormalization group .
the effective couplings obey a standard renormalization group first - order differential equation controlled by the anomalous - dimension matrix .
it is clear that the only information that can be extracted from the effective lagrangian , without knowing the full theory , is this anomalous - dimension matrix .
however , to obtain the couplings it is necessary to solve the renormalization group equations , and this requires the knowledge of @xmath273 boundary conditions ( if @xmath273-operators are involved in the mixing ) .
one - loop calculations with the effective lagrangian can only be used to relate the couplings of operators at different scales , but give no information at all on the actual value of those couplings . only experiment , or a more complete theory , can give new information on the values of the effective lagrangian couplings .
this trivial observation is of importance when using the effective lagrangian approach to classify the sort of new physics one could find in future experiments .
after renormalization we obtain an effective lagrangian that can be split in three pieces . the first one , @xmath39 , is just the standard model lagrangian ( expressed in terms of effective couplings ) ; the second one is the four - lepton interaction obtained from the full theory at tree level , @xmath158 ; the third piece , @xmath252 contains all dimension - six operators generated at the one - loop level in the full model .
this effective lagrangian was rewritten by using the standard model classical equations of motion in order to display its physical content more transparently .
some caution is needed when using the equations of motion , especially for operators that are going to be inserted in loop diagrams . in our case , however , the equations of motion are used only in operators that are generated at one loop in the full theory and that are supposed to be used only at tree level in the effective lagrangian . then , the use of equations of motion is completely legitimate .
we used them for the unbroken theory , but it is worth while to stress that the equations of motion might also be used after ssb ( or might not be used at all ) and this would not change any physical result .
we also discussed the consequences of ssb on our effective lagrangian by substituting the vev of the doublet and neglecting all higgs interactions .
apart from a negligible modification of the relation between the masses of the vector bosons , the most interesting consequence of the model is due to the different operators contributing to processes with violation of the generational lepton numbers .
we made a short review of some processes generated by our one - loop effective lagrangian , @xmath274 , @xmath275 , etc ( where @xmath276 are different flavour indices ) .
a more detailed phenomenological analysis will be presented elsewhere . finally , to see how one can use our effective lagrangian ,
we have calculated the decay width of the @xmath0 gauge boson to a pair of different leptons .
it contains contributions from one - loop diagrams with one insertion of the tree - level four - fermion operator and direct contributions from operators generated at the one - loop level .
we have done also the calculation in the full theory and compared the results . for a mass of the charged scalar @xmath61 larger than the @xmath0 mass
, the effective theory calculation gives a good approximation .
we thank j. bernabu , a. casas , s. fanchiotti , s. peris and a. pich for helpful discussions on the subject of this paper and for a critical reading of the manuscript .
this work was supported in part by cicyt , spain , under grant aen93 - 0234 . * appendices *
[ det ] in this appendix we calculate the effective lagrangian contribution coming from the determinant of a fluctuation operator of the form @xmath277 .
it is given by @xmath278 where @xmath140 is the covariant derivative for a generic gauge group , for example an @xmath279 , and @xmath280 is a general matrix valued function of @xmath281 .
powerful methods to calculate traces of these kinds of operators have been developed in the last years @xcite .
we will employ some techniques developed in all of those papers , but we will follow more closely those in ref . @xcite .
we will use the following notation @xmath282 where @xmath283 are the generators of the gauge group .
they satisfy the normalization condition @xmath284 .
then , the covariant derivative acts on @xmath280 as follows : @xmath285 \ , \ ] ] the field strength tensor is defined as @xmath286 \ .\ ] ] we will use the metric @xmath287 and the following conventions for the momentum operator and plane wave states : @xmath288 the normalization of the states is ( in @xmath156 dimensions ) @xmath289 and @xmath290 and @xmath291 .
let @xmath292 be an operator ; then we understand the trace of that operator to be @xmath293 where `` tr '' means trace over all degrees of freedom , space - time and internal , while tr " is the trace over only the internal degrees of freedom
. then @xmath294 here @xmath295 is the operator @xmath292 in the representation of positions @xmath296 . in our case
we have @xmath297 by using eq .
( [ tro ] ) and the operator identity @xmath298 , we find @xmath299 where the factor @xmath300 at the end indicates that the operators act on the identity .
now we can expand the logarithm in the following form @xmath301 the first term is the usual coleman - weinberg term . it is a constant term , which only contributes to the energy density , and we will drop it . comparing eq .
( [ effprel ] ) with the expression for the effective lagrangian , eq .
( [ ap1 ] ) , we find @xmath302 it is clear that eq .
( [ lexpansion1 ] ) gives an expansion in powers of @xmath84 @xmath303 where @xmath304 are traces of gauge - invariant operators of dimension @xmath305 built from @xmath230 and @xmath306 ( and their covariant derivatives ) .
in general @xmath307 here @xmath308 is a linearly independent set of traces of operators of dimension @xmath305 ( operators that can be related using partial integration of @xmath140 , cyclic permutations inside the trace , or bianchi identities are not considered to be linearly independent ) .
we use the following basis for these traces of gauge - invariant operators : @xmath309 the normalization of the @xmath304 is such that @xmath310 all the integrals will be dimensionally regularized and @xmath163 is the dimensional - regularization scale . as we will see , with this normalization the coefficients @xmath311 will be just numbers and independent from @xmath156 . then
, our task is to evaluate the coefficients @xmath311 .
they could be evaluated directly by expanding eq .
( [ lexpansion1 ] ) , performing the momentum integrals and using integration by parts to rewrite it in a canonical form . however , it is better to use another method @xcite .
the trick is as follows : as we know that eqs .
( [ lexpansion1 ] ) , ( [ lexpansion2 ] ) and ( [ lop1 ] ) are valid for any @xmath230 and any @xmath306 , with the @xmath311 independent from @xmath230 and @xmath306 , we can compute @xmath311 using a particular configuration of @xmath230 and @xmath306 . then , the resulting @xmath311 will be valid for any @xmath230 and @xmath306 . as we want to compute ( [ lexpansion1 ] ) and we have @xmath312 ,
it is obvious that if we choose @xmath313 and @xmath314 we will have @xmath315 .
thus , to calculate eq .
( [ lexpansion1 ] ) we will choose the following ( constant ) configuration for the field @xmath230 : @xmath316 to avoid confusion we will denote the field @xmath230 as @xmath317 in this special configuration . in this configuration
we obviously have @xmath318,{\hspace{1cm}}d_\mu g = [ n_\mu , g]\ , \ ] ] where @xmath319 is any matrix valued function of @xmath230 and @xmath306 in the special configuration ( hence , constant with respect to space - time variables ) .
the important point is that eq .
( [ specialconf ] ) can be inverted , thus allowing us to pass from the special configuration to the general configuration . since in the special configuration ( [ specialconf ] ) we have @xmath320 eq .
( [ lexpansion1 ] ) reads @xmath321 where we have taken into account that integrals with an odd number of @xmath121 s cancel under symmetric integration and we have redefined @xmath322 .
now we can use that @xmath323 here @xmath324 is the completely symmetric tensor with @xmath305 indices built only with the metric tensor @xmath325 .
for instance @xmath326 and @xmath327 . in general
it contains @xmath328 terms . inserting eq .
( [ integsym ] ) in eq .
( [ lagconspe1 ] ) , we finally obtain @xmath329 comparing this result with eq .
( [ lexpansion2 ] ) and eq .
( [ coeff1 ] ) we find that in the special configuration ( [ specialconf ] ) we have @xmath330 where @xmath331 represents the sum of all possible products of @xmath305 @xmath317 s contracted pairwise .
for instance , @xmath332 and @xmath333 .
traces of different operators can be related by using the cyclic property of the trace .
we find for example @xmath334 .
then , in the special configuration we have @xmath335 where @xmath336
are traces of strings of @xmath317 s contracted pairwise and the coefficients @xmath337 include the normalization factor @xmath338 .
if we define the set of linearly independent traces built from @xmath317 as follows : @xmath339 we find @xmath340 but we know from eq .
( [ lop1 ] ) that in the general case @xmath304 can be expressed as a linear combination of traces of gauge - invariant operators @xmath308 .
then , in the special configuration we can write @xmath304 in terms of the two sets of operators @xmath341 but in the special configuration we can relate the operators @xmath308 with the operators @xmath336 by using eq .
( [ specialconf ] ) . in general we will have @xmath342 the important point of these equations
is that the above linear transformation @xmath343 can can be inverted .
this will allow us to compute the @xmath344 s in terms of the already known @xmath345 s .
in matrix form we obtain @xmath346 where @xmath347 is the transposed matrix of @xmath343 .
we skip the details of a rather tedious calculation and quote only the final results valid for an arbitrary configuration @xmath348 these results agree with the results obtained in @xcite by using proper time methods ( after passing to minkowski space ) .
the effective lagrangian obtained from the determinant of the fluctuation operator is given by eq .
( [ lexpansion2 ] ) with the @xmath304 given above and with the @xmath349 given in eq .
( [ coeff1 ] ) . for @xmath164 and
in the limit @xmath350 we obtain @xmath351 where @xmath352 contains the divergent part when @xmath353 and only appears in the first two coefficients .
[ heavy - light ] here we collect the relevant amplitudes needed to obtain the effective operators generated by one - loop diagrams in the full theory with heavy - light particles in the loops .
we have always used dimensional regularization with anticommuting @xmath354 . from the self - energy diagrams in the full theory ( fig .
3 ) we find @xmath355 where @xmath356 is the left - handed chirality projector . in the effective theory , the doublet self - energy diagram with the four - fermion coupling ( fig .
3.b ) is zero in dimensional regularization because it is a massless tadpole - like diagram . the diagrams of figs .
4.a and 4.b give the following amplitudes for the coupling of the @xmath109 to the lepton doublet @xmath357 here @xmath358 , and we have kept all external momenta off - the - mass - shell . in the diagram
4.a we split the charged scalar propagator as described in section [ matching ] .
the result given here corresponds to the difference between the full theory diagram ( fig .
4.a ) and the vertex obtained in the effective theory with the four - fermion interaction ( fig .
the full amplitude for the @xmath109 vertex is @xmath359 for the vertex of the @xmath138 only the diagram 4.a exists since it does not couple to the @xmath20 singlet scalar .
the result is @xmath360 from these amplitudes together with the self - energy amplitudes , one can easily reconstruct the gauge - invariant operators that generate them as done in section [ matching ] . from the diagram of fig .
5.b we obtain the amplitude @xmath361 while for the diagram of fig .
5.a we get @xmath362 here @xmath363 and @xmath364 are the momenta of an incoming @xmath365 and an outgoing @xmath365 respectively , and @xmath161 is defined in eq .
( [ defhatf ] ) . as always
, we have used the splitting of the charged scalar propagator @xmath366 and have only computed the second term .
the first term gives rise to the effective lagrangian contribution given by the diagram of fig .
5.c . from eq .
( [ amp5b ] ) we obtain the following effective operators @xmath367 which can be written in the form of eq .
( [ op5b ] ) after using the following @xmath20 fierz transformations : @xmath368 here @xmath300 is the @xmath203 identity matrix . the amplitude corresponding to the box with the two charged scalars in the loop ( fig .
6.c and crossed ) has the following form @xmath369
[ fcdecay ] in this appendix we give some results of the calculation of the decay width for @xmath36 done with both the full and the effective lagrangians .
we show in fig .
4.c the diagram responsible for this decay in the effective theory ( the wavy line is , in this case , a @xmath0 gauge boson , the external leptons are charged leptons or neutrinos , and the lines in the loop represent neutrinos or charged leptons , respectively ) .
the corresponding amplitudes are ( for massless external leptons and all particles on - the - mass - shell ) : @xmath370 to this amplitude we should add the contribution coming from the operators generated through matching , eq .
( [ fczecoupling ] ) .
the total result is @xmath371 for the @xmath0 decay to neutrinos , a similar calculation gives the following result : @xmath372 in both cases the divergent term coming from the loop integration has been cancelled by the one - loop effective operator contribution .
previous results should be compared with the full lagrangian amplitudes , @xmath373 where , using the variable @xmath374 we have @xmath375 and @xmath376 here @xmath377 is the dilogarithmic function . the functions @xmath265 and @xmath266 have the following asymptotic values for @xmath268 :
@xmath378 these results are in complete agreement with the effective lagrangian calculation .
in fact the logarithmic contribution coming from @xmath265 can easily be obtained from the calculation of diagram 4.c .
however , the non - logarithmic part of that amplitude is arbitrary , in fact divergent .
the only way to fix it is by matching the full theory
. 10 , phys .
. lett . * 18 * ( 1967 ) 188 , phys . rev .
* 166 * ( 1968 ) 1568 , phys .
* 177 * ( 1969 ) 2239 , in lectures on elementary particles and quantum field theory , ed .
s. deser , m. grisaru and h. pendleton ( mit , cambridge , 1970 ) , physica * 96a * ( 1979 ) 327 , phys .
* 43 * ( 1979 ) 1562 , phys .
lett . * 91b * ( 1980 ) 51 , phys . rev .
* d21 * ( 1980 ) 3369 , phys . rev .
* d25 * ( 1982 ) 1605 , annals of physics * 158 * ( 1984 ) 142 , nucl . phys .
* b250 * ( 1985 ) 517 , nucl . phys . *
b361 * ( 1991 ) 339 , naturw .
* 23 * ( 1935 ) 246 , ann .
* 26 * ( 1936 ) 398 , z. phys .
* 98 * ( 1936 ) 714 , ric
. scient . * 4 * ( 1934 ) 491 , nuovo cim .
* 11 * ( 1934 ) 1 , z. phys .
* 88 * ( 1934 ) 161 , weak interactions and modern particle theory , ( benjamin / cummings , menlo park , 1984 ) , phys
b232 * ( 1989 ) 113 , phys . lett . *
b237 * ( 1990 ) 527 , phys .
* b4 * ( 1971 ) 3174 , phys .
* 12 * ( 1974 ) 75 , nucl
b231 * ( 1984 ) 269 , effective field theory and the fermi surface , nsf - itp-92 - 132 , lectures at tasi92 , boulder , 1992 , conference summary , uttg-25 - 92 , talk presented at the xxvi international conference on high energy physics , dallas , texas , august , 1992 , phys .
* d11 * ( 1975 ) 2856 , phys . rev .
* d22 * ( 1980 ) 200 , phys . rev .
* d22 * ( 1980 ) 1166 , nucl . phys . *
b188 * ( 1981 ) 118 , nucl . phys .
* b297 * ( 1988 ) 183 , phys . lett . *
b255 * ( 1991 ) 405 , nucl . phys . *
b373 * ( 1992 ) 117 , in supergravity , proceedings of the stony brook workshop , stony brook , new york , 1979 , ed .
p. van nieuwenhuizen and d. freedman ( north - holland , amsterdam , 1980 ) , in proceedings of the workshop on unified theories and baryon number in the universe , tsukuba , japan , 1979 , ed .
a. sawada and a. sugamoto ( kek report no .
79 - 18 , tsukuba , 1979 ) , phys .
* 29 * ( 1972 ) 669 , phys
* 43 * ( 1979 ) 1566 , phys .
* 43 * ( 1979 ) 1571 , nucl
b173 * ( 1980 ) 269 , nucl . phys .
* b268 * ( 1986 ) 621 , phys . lett . *
b265 * ( 1991 ) 326 , nucl .
* b228 * ( 1983 ) 464 , z. phys . *
c31 * ( 1986 ) 433 , trilinear couplings among the electroweak vector bosons and their determination at lep2 , bi - tp 92/44 ( to appear in nucl . phys . ) , 1992 , z. phys . * c59 * ( 1993 ) 133 , nucl
b384 * ( 1992 ) 3 , phys .
b307 * ( 1993 ) 116 , phys . lett .
* b283 * ( 1992 ) 353 , low energy effects of new interactions in the electroweak boson sector , mad / ph/737 , 1993 , z. phys . * c1 * ( 1979 ) 259 , nucl . phys . *
b282 * ( 1987 ) 253 , uses and abuses of effective lagrangians , mcgill-92 - 05 ( revised version , june 1993 ) , 1992 , phys .
* 69 * ( 1992 ) 3428 , effective lagrangian approach to precision measurements : the anomalous magnetic moment of the muon , nsf - itp-92 - 122 , 1992 , beyond the standard model with effective lagrangians , slac - pub-12 , 1993 , effective lagrangians and triple boson couplings , ucrhep-108 , 1993 , phys
* d22 * ( 1980 ) 2860 , phys . lett . *
93b * ( 1980 ) 389 , nucl .
b248 * ( 1984 ) 59 , nucl . phys . *
b248 * ( 1984 ) 77 , phys .
* 59 * ( 1987 ) 385 , phys
d44 * ( 1991 ) 2139 , heavy top quark effects in vertices with light external fermions , um - th-93 - 05 , 1993 , nucl
b387 * ( 1992 ) 523 , phys . lett . *
115b * ( 1982 ) 401 , phys . lett .
* 162b * ( 1985 ) 153 , nucl .
b310 * ( 1988 ) 714 , phys .
* d43 * ( 1991 ) 2337 , nucl
b357 * ( 1991 ) 222 , nucl .
b104 * ( 1976 ) 445 , nucl . phys . * b122 * ( 1977 ) 109 , nucl .
b178 * ( 1981 ) 75 , nucl . phys .
* b172 * ( 1980 ) 349 , reduced effective lagrangians , um - th-92 - 28 , 1992 , phys .
* d31 * ( 1985 ) 2605 , phys . rev .
* 54 * ( 1985 ) 1222 , phys
* d32 * ( 1985 ) 2653 , phys . rev .
* d33 * ( 1986 ) 3645 , phys . rev .
* 57 * ( 1986 ) 1199 , nucl
b268 * ( 1986 ) 669 , phys .
* 182 * ( 1989 ) 1 , nucl .
b62 * ( 1973 ) 444
* figure 1 : * diagram a ) is the tree - level diagram contributing to the effective lagrangian .
the solid lines represent lepton doublets and the thick dashed line represents the heavy scalar .
diagram b ) represents the four - lepton interaction in the effective theory .
the symbol @xmath379 means a scalar current insertion .
0.4cm*figure 2 : * diagrammatic representation of the contributions to the one - loop effective lagrangian given by the determinant of the fluctuation operator .
gauge bosons are represented by wavy lines and the standard higgs is represented by thin dashed lines .
0.4cm*figure 3 : * matching conditions : self - energy diagrams .
a ) in the full theory and b ) in the effective one . 0.4cm*figure 4 : * matching conditions : penguin diagrams a ) and b ) in the full theory and c ) in the effective one . 0.4cm*figure 5 :
* matching conditions : seagull diagrams a ) and b ) in the full theory and c ) in the effective one . 0.4cm*figure 6 : * matching conditions : box diagrams a ) and b ) in the full theory and c ) in the effective one .
0.4cm*figure 7 : * the family lepton - number violating branching ratio @xmath380 computed in both the full ( solid line ) and the effective ( dashed line ) theories as a function of the charged scalar mass .
the couplings @xmath381 are taken to be equal to 1 .
|
we study several problems related to the construction and the use of effective lagrangians by considering an extension of the standard model that includes a heavy scalar singlet coupled to the leptonic doublet .
starting from the full renormalizable model , we build an effective field theory by integrating out the heavy scalar . a local effective lagrangian ( up to operators of dimension six )
is obtained by expanding the one - loop effective action in inverse powers of the heavy mass .
this is done by matching some green functions calculated with both the full and the effective theories . using this simple example we study the renormalization of effective lagrangians in general and
discuss how they can be used to bound new physics .
we also discuss the effective lagrangian after spontaneous symmetry breaking , and the use of the standard model classical equations of motion to rewrite it in different forms .
the final effective lagrangian in the physical basis is wel suited to the study of the phenomenology of the model , which we comment on briefly . finally ,
as an example of the use of our effective field theory , we consider the leptonic flavour - changing decay of the @xmath0 boson in the effective theory and compare the results obtained with the full model calculation .
-1 in -1 in 16.3 cm to to 0.3em department of physics + univ .
bielefeld , postfach 8640 + d-33615 bielefeld ,
germany0.5em 0.3em and0.5em 0.3em th division , cern , 1211 genve 23 , switzerland0.5em cern - th.7030/93 + october 1993 ) alexander von humboldt fellow . on leave of absence from the joint institute for nuclear research , dubna , russia . * ) on leave of absence from departament de fsica terica , universitat de valncia , and ific , valncia , spain .
|
it has long been the goal of mathematicians to minimize the set of assumptions and axioms in their systems .
implementers of theorem provers use this principle : they use a logic with as few inference rules as possible , and prove lemmas outside the core logic in preference to adding new inference rules . in applications of logic to computer security such as _ proof - carrying code _
@xcite and distributed authentication frameworks @xcite the implementation of the core logic is inside the trusted code base ( tcb ) , while proofs need not be in the tcb because they can be checked .
two aspects of the core logic are in the tcb : a set of logical connectives and inference rules , and a program in some underlying programming language that implements proof checking that is , interpreting the inference rules and matching them against a theorem and its proof .
definitions and lemmas are essential in constructing proofs of reasonable size and clarity .
a proof system should have machinery for checking lemmas , and applying lemmas and definitions , in the checking of proofs .
this machinery also is within the tcb ; see figure [ lemmamach ] .
[ lemmamach ] many theorem provers support definitions and lemmas and provide a variety of advanced features designed to help with tasks such as organizing definitions and lemmas into libraries , keeping track of dependencies , and providing modularization ; in our work we are particularly concerned with separating that part of the machinery necessary for proof checking ( i.e. , in the tcb ) from the programming - environment support that is used in proof development .
this separation was particularly important for a proof - carrying code system we built initially in @xmath0prolog @xcite . in this paper
we will demonstrate a definition / lemma implementation that is about three dozen lines of code .
the @xmath0prolog language @xcite has several features that allow concise and clean implementation of logics , proof checkers , and theorem provers @xcite . in a previous paper @xcite
, we presented a lemma and definition mechanism implemented in @xmath0prolog . in this paper , we extend that work and describe it more fully .
we present the lemma mechanism and a generalization of our definition mechanism , again implemented in @xmath0prolog .
since we now have more experience using the twelf system @xcite , we include a detailed comparison of the twelf and @xmath0prolog versions of the encoding of our logic , lemmas , and definitions .
an important purpose of this paper is to show which language features allow a small tcb and efficient representation of proofs . we also give a comparison of programming issues that are important to our proof - carrying code application .
although the lemma and definition mechanism is general , we illustrate it using an implementation of higher - order logic .
we call this logic the _ object logic _ to distinguish it from the _ metalogic _ implemented by @xmath0prolog or twelf .
our object logic is not polymorphic , but our lemma and definition mechanisms are polymorphic in the sense that they can express properties that hold at any type of the object logic .
the symmetry of equality , for example , is one such lemma we will encounter .
the @xmath0prolog version of the clauses we present use the syntax of the terzo implementation @xcite .
we also discuss the teyjus implementation @xcite and compare the two for our purposes .
terzo is interpreted and provides more flexibility , but teyjus has a compiler in which our code runs much more efficiently .
@xmath0prolog is a higher - order logic programming language which extends prolog in essentially two ways .
first , it replaces first - order terms with the more expressive simply - typed @xmath0-terms ; @xmath0prologimplementations generally extend simple types to include ml - style prenex polymorphism @xcite .
second , it permits implication and universal quantification ( over objects of any type ) in goal formulas .
we introduce types and constants using ` kind ` and ` type ` declarations , respectively .
for example , a new primitive type @xmath1 and a new constant @xmath2 of type @xmath3 are declared as follows .
.... kind t type .
type f t - > t - > t. .... capital letters in type declarations denote type variables and are used in polymorphic types . in program goals and clauses , @xmath0-abstraction is written using backslash 134 as an infix operator .
capitalized tokens not bound by @xmath0-abstraction denote free variables .
all other unbound tokens denote constants .
universal quantification is written using the constant pi in conjunction with a @xmath0-abstraction ( e.g. , ` pi x134 ` represents universal quantification over variable x ) .
the symbols _ comma _ and = > represent conjunction and implication .
the symbol : - denotes the converse of = > and is used to write the top - level implication in clauses .
the type ` o ` is the type of clauses and goals of @xmath0prolog . we usually omit universal quantifiers at the top level in definite clauses , and assume implicit quantification over all free variables .
we will encode a natural deduction proof system for our higher - order object logic .
( in our earlier work @xcite , we implemented a sequent calculus version . )
we implement a proof checker for this logic that is similar to the one described by felty @xcite .
program [ core - types ] contains the type declarations used in our encoding . ....
kind tp type .
kind tm type .
kind pf type .
type form tp .
type intty tp .
type arrow tp - > tp - > tp .
infixr arrow 8 .
type pair tp - > tp - > tp .
type eq tp - > tm - > tm - > tm .
type imp
tm - > tm - > tm .
infixr imp 7 .
type forall tp - > ( tm - > tm ) -
> tm . type false tm .
type lam ( tm - > tm ) -
> tm . type app tp - > tm - > tm - > tm .
type mkpair tm - > tm - > tm .
type fst tp - > tm - > tm .
type snd tp - > tm - > tm .
type hastype tm - > tp - > o. type proves pf - > tm - > o. type assump o - > o. type refl pf .
type beta pf .
type fstpair pf .
type sndpair pf .
type surjpair pf .
type congr tp - > tm - > tm - > ( tm - > tm ) - > pf - > pf - > pf .
type imp_i ( pf - > pf ) - > pf .
type imp_e tm - > pf - > pf - > pf .
type forall_i ( tm - > pf ) - > pf .
type forall_e tp - > ( tm - > tm ) - > pf - > tm - > pf . .... [ core - types ] we introduce three primitive types : ` tp ` for object - level types , ` tm ` for object - level terms ( including formulas ) and ` pf ` for proofs in the object logic .
we introduce constants for the object - level type constructors .
the main type constructor for our object language is the ` arrow ` constructor taking two types as arguments .
we also include objects of type ` tp ` to represent base types , such as ` form ` and ` intty ` .
to represent formulas , we introduce constants such as ` imp ` to represent implication in the object logic , and ` eq ` which takes two terms and a type and is used to represent equality at any type .
we use infix notation for the type arrow and binary logical connectives .
the binding strength of each infix operator is declared using an ` infix ` declaration .
the constant ` forall ` represents universal quantification .
it takes a type representing the type of the bound variable and a functional argument , which allows object - level binding of variables by quantifiers to be defined in terms of meta - level @xmath0-abstraction .
an example of its use is the following formula , which expresses the commutativity of equality for integers : .... forall intty ( x\ forall intty ( y\ ( eq intty x y ) imp ( eq intty y x ) ) ) .
.... the parser uses the usual rule for the syntactic extent of a lambda , so this expression is equivalent to .... forall intty x\ forall intty y\ eq intty x y imp eq intty y x. ....
this use of higher - order data structures is called _ higher - order abstract syntax _ @xcite ; with it , we do nt need to describe the mechanics of substitution explicitly in the object logic @xcite . to represent terms
, we introduce the ` app ` and ` lam ` constants for application and abstraction , as well as constants for pairing and projections .
the ` app ` constructor takes three arguments .
the second argument is a term of functional type and the third argument is the term it is applied to .
the first argument is the type of the argument to the function .
the ` lam ` constant has a type , which like ` forall ` , uses meta - level abstraction to represent object - level binding .
the constants at the end of program [ core - types ] are used to build terms representing proofs .
we call these constants as well as any other terms whose type ends in `` ` - > pf ` '' _ proof constructors_. programs [ core - types ] and [ core - rules ] together implement a full proof checker for our object logic
. .... hastype ( eq t x y ) form : - hastype x t , hastype y t. hastype ( a imp b ) form : - hastype a form , hastype b form .
hastype ( forall t a ) form : - pi x\ ( hastype x t = > hastype ( a x ) form ) .
hastype false form .
hastype ( lam f ) ( t1 arrow t2 ) : - pi x\ ( hastype x t1 = > hastype ( f x ) t2 ) . hastype ( app t1 f x ) t2 : - hastype f ( t1 arrow t2 ) , hastype x t1 .
hastype ( mkpair x y ) ( pair t1 t2 ) : - hastype x t1 , hastype y t2 .
hastype ( fst t2 x ) t1 : - hastype x ( pair t1 t2 ) .
hastype ( snd t1 x ) t2 : - hastype x ( pair t1 t2 )
. proves q a : - assump ( proves q a ) .
proves refl ( eq t x x ) . proves beta ( eq t2 ( app t1 ( lam f ) x ) ( f x ) ) .
proves fstpair ( eq t1 ( fst t2 ( mkpair x y ) ) x ) .
proves sndpair ( eq t2 ( snd t1 ( mkpair x y ) ) y ) . proves surjpair ( eq ( pair t1 t2 ) ( mkpair ( fst t2 z ) ( snd t1 z ) ) z ) . proves ( congr t x z h p1
p2 ) ( h x ) : - hastype
x t , hastype z t , proves p1 ( eq t x z ) , proves p2 ( h z ) .
proves ( imp_i q ) ( a imp b ) : - pi p\ ( assump ( proves p a ) = > proves ( q p ) b )
. proves ( imp_e a q1 q2 ) b : - hastype a form , proves q1 ( a imp b ) , proves q2 a. proves ( forall_i q ) ( forall t a ) : - pi y\ ( hastype y t = > proves ( q y ) ( a y ) )
. proves ( forall_e t a q x ) ( a x ) : - pi x\ ( hastype x t = > hastype ( a x ) form ) , hastype x t , proves q ( forall t a ) . ....
[ core - rules ] program [ core - rules ] implements both typechecking and inference rules .
the last four clauses of program [ core - rules ] implement the introduction and elimination rules for implication and universal quantification , which are given in figure [ core - inf ] .
[ core - inf ] we do not include inference rules for the other logical connectives . instead , we define them in terms of existing connectives using our definition mechanism described later .
the remaining clauses for the ` proves ` predicate implement inference rules for equality .
typechecking for terms is implemented by the ` hastype ` clauses .
proof checking is implemented by the ` proves ` clauses .
a goal of the form ` ( proves p a ) ` should be run only after ` a ` is typechecked , i.e. , a proper check has the form ` ( hastype a form , proves p a ) ` . to implement the discharge of assumptions in the implication introduction rule , we use implication and universal quantification in @xmath0prolog goals .
the goal ` ( d = > g ) ` adds clause ` d ` to the @xmath0prolog clause database , attempts to solve ` g ` , and then ( upon either the success or failure of ` g ` ) removes ` d ` from the clause database .
the goal ` ( pi y134(g y ) ) ` introduces a new constant ` c ` with the same type as ` y ` , replaces ` y ` with ` c ` , and attempts to solve the goal ` ( g c ) ` .
for example , consider the goal .... proves ( imp_i q\q ) ( a imp a ) .... where ` a ` is a propositional constant ( a constant of type ` form ` ) ; then @xmath0prolog will execute the ( instantiated ) body of the ` imp_i ` clause .... pi p\ ( assump ( proves p a ) = > proves ( ( q\q ) p ) a ) .... this generates a new constant ` c ` , and adds ` ( assump ( proves c a ) ` to the database ; then the subgoal ` ( proves ( ( q134q ) c ) a ) ` , which is @xmath4-equivalent to ` ( proves c a ) ` , matches the first clause for the ` proves ` predicate .
the subgoal ` ( assump ( proves c a ) ) ` is generated and this goal matches our dynamically added clause .
we have chosen to use the ` assump ` predicate for adding atomic clauses to the program .
this is not necessary , but we find it useful to distinguish between adding atomic clauses and adding non - atomic clauses , which we will see later .
note that the typechecking clauses for ` forall ` and ` lam ` use meta - level implication and universal quantification in a manner similar to the ` proves ` clause for the @xmath5-i rule .
it is important to show that our encoding of higher - order logic in @xmath0prolog is _
adequate_. to do so , we must show that a formula has a natural deduction proof if and only if its representation as a term has an associated proof term that can be checked using the inference rules of program [ core - rules ] .
the encoding we use is similar to the encoding of higher - order logic in the logical framework @xcite and the proof of adequacy of our encoding is similar to the one discussed there .
the main difference between the two encodings is the types of the logical connectives .
for example , in their encoding , ` imp ` is given type ` tm ` and the fact that it is a connective which takes two formulas as arguments is expressed using object level types ; the ` hastype ` clause is .... hastype imp ( form arrow form arrow form ) .
.... an implication must then be expressed using the ` app ` constructor , e.g. , ` ( app ( app imp a ) b ) ` .
we found that this encoding of the connectives quickly became cumbersome and our encoding was more readable . on the other hand ,
our encoding is not as economical as the one we used previously @xcite .
there we represented object - level types as meta - level types , which allowed us to eliminate all the ` hastype ` clauses and subgoals .
the types of our object logic , however , did not match up well with the types of @xmath0prolog , which forced certain limitations in the implementation of our proof - carrying code system .
( see appel and felty @xcite for further analysis . )
the encoding in the current paper seems to be the best compromise .
in mathematics the use of lemmas can make a proof more readable by structuring the proof , especially when the lemma corresponds to some intuitive property . for automated proof checking ( in contrast to automated or traditional theorem proving )
this use of lemmas is not essential , because the computer does nt need to understand the proof in order to check it . but
lemmas can also reduce the _
size _ of a proof ( and therefore the time required for proof checking ) : when a lemma is used multiple times it acts as a kind of `` subroutine . ''
this is particularly important in applications like proof - carrying code where proofs are transmitted over networks to clients who check them .
we first present an example which we use to illustrate our lemma mechanism in @xmath0prolog ( section [ lemmasexample ] ) , and then present this mechanism as we d implement it in terzo ( section [ lemmasterzo ] ) .
we then explain the modifications required to meet the extra restrictions imposed by teyjus ( section [ lemmasteyjus ] ) .
we end this section with some optimizations that are important for keeping proofs that use lemmas as small as possible ( section [ lemmasopt ] ) and then with some more examples ( section [ lemmasmore ] ) . theorem [ thm : simple ] shows the use of our core logic to express a simple proof checking goal .
.... proves ( forall_i i\ ( forall_i j\ ( imp_i q\ ( congr intty i j ( eq intty j ) q refl ) ) ) ) ( forall intty i\ forall intty j\ ( eq intty
i j imp eq intty j i ) )
. .... [ thm : simple ] the proof of this lemma uses the @xmath6-i rule as well as congruence and reflexivity of equality .
its proof can be checked as a successful @xmath0prolog query to our core logic in programs [ core - types ] and [ core - rules ] .
alternatively , we may want to prove it using the following general lemma about symmetry of equality at any type .
@xmath7 the proof of this lemma can be checked as the following @xmath0prolog query . ....
pi t\ pi a\ pi b\ pi p\ ( hastype a t , hastype b t , proves p ( eq t b a ) ) = > proves ( congr t b a ( eq t a ) p refl ) ( eq t a b ) . ....
this query introduces an arbitrary ` p ` , adds the typing clauses ` ( hastype a t ) ` and ` ( hastype b t ) ` , and the assumption ` ( proves p ( eq t b a ) ) ` to the set of clauses , then checks the proof of congruence using these facts .
the syntax @xmath8 ` = > ` @xmath9 means exactly the same as @xmath9 ` : - ` @xmath8 , so we could just as well write this query as .... pi t\ pi a\
pi b\ pi p\ ( proves ( congr t b a ( eq t a ) p refl ) ( eq t a b ) : - hastype a t , hastype b t , proves p ( eq t b a ) ) . .... now , suppose we abstract the proof ( roughly , ` congr t b a ( eq t a ) p refl ` ) from this query . ....
( inference = ( pcon\ pi t\ pi a\ pi b\ pi p\ proves ( pcon t a b p ) ( eq t a b ) : - hastype a t , hastype b t , proves p ( eq t b a ) ) , proof = ( t\a\b\p\ congr t b a ( eq t a ) p refl ) , query = ( inference proof ) , query ) . ....
the solution of this query proceeds in four steps : the variable ` inference ` is unified with a @xmath0-term ; ` proof ` is unified with a @xmath0-term ; ` query ` is unified with the application of ` inference ` to ` proof ` ( which is a term @xmath4-equivalent to the query of the previous paragraph ) , and finally ` query ` is solved as a goal ( checking the proof of the lemma ) .
once we know that the lemma is valid , we can make a new @xmath0prolog atom ` symm ` to stand for its proof , and we can prove some other theorem in a context where the clause ` ( inference symm ) ` is in the clause database ; remember that ` ( inference symm ) ` is @xmath4-equivalent to .... pi t\ pi a\ pi
b\ pi p\ ( proves ( symm t a b p ) ( eq t a b ) : - hastype a t , hastype b t , proves p ( eq t b a ) ) . .... this series of transformations starting with a proof checking subgoal has led us to a clause that looks remarkably like an inference rule . with this clause in the database , we can use the new proof constructor ` symm ` just as if it were primitive . instead of adding new clauses like this to our proof checker , which would increase the size of our tcb ,
we show how to put such lemmas inside proofs . in the example in the previous section , `
symm ` is a new constant , but when lemmas are proved and put inside proofs dynamically , we can instead `` make a new atom '' by simply ` pi`-binding it .
this leads to the recipe for lemmas shown in program [ lemmafig ] , which is the heart of our lemma mechanism . ....
type lemma_pf ( a - > o ) - > a - > ( a - > pf ) - > pf .
proves ( lemma_pf inference lemmaproof restproof ) c : - inference lemmaproof , pi name\ ( ( inference name ) = > ( proves ( restproof name ) c ) ) .
.... [ lemmafig ] ( we will improve it slightly in the next section . ) this program introduces a constructor ` lemma_pf ` for storing lemmas in proofs .
this constructor takes three arguments : ( 1 ) a derived inference rule ` inference ` ( of type ` a - > o ` ) parameterized by a proof constructor ( of type ` a ` ) , ( 2 ) a term ` lemmaproof ` of type ` a ` representing a proof of the lemma built from core - logic proof constructors ( or using other lemmas ) , and ( 3 ) a proof of the main theorem ` restproof ` that is parameterized by a proof constructor ( of type ` a ` ) .
operationally , this clause first executes ` ( inference lemmaproof ) ` as a query , to check the proof of the lemma itself ; then it ` pi`-binds ` name ` in the lemma , adds it as a new clause , and runs ` restproof ` ( which is parameterized on the lemma proof constructor ) applied to ` name ` .
the terms ` inference ` and ` proof ` from the example in section [ lemmasexample ] illustrate the form of the terms which will appear as the first two arguments to ` lemma_pf ` .
theorem [ thm : symm ] illustrates the use of ` lemma_pf ` in an example ; this theorem is a modification of theorem [ thm : simple ] that uses the ` symm ` lemma .
.... proves ( lemma_pf ( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm t a b p ) ( eq t a b ) : - hastype a t , hastype b t , proves p ( eq t b a ) ) ( t\a\b\p\ ( congr t b a ( eq t a ) p refl ) ) ( symm\ ( forall_i i\ ( forall_i j\ ( imp_i q\ ( symm intty j i q ) ) ) ) ) ) ( forall intty i\ forall intty j\ ( eq intty i j imp eq intty j i ) ) . .... [ thm : symm ] if we restrict ourselves to the terzo implementation of @xmath0prolog , then meta - level formulas can occur inside proofs using any of the @xmath0prolog connectives .
but if we want to be able to use teyjus as well , we must make one more change .
the teyjus system does not allow ` = > ` or ` : - ` to appear in arguments of predicates .
thus the term .... ( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm t a b p ) ( eq t a b ) : - hastype a t , hastype b t , proves p ( eq t b a ) ) .... occurring in the ` symm ` lemma in theorem [ thm : symm ] can not appear directly as the first argument to ` lemma_pf ` .
teyjus also does not allow variables to appear at the head of the left of an implication .
these restrictions come from the theory underlying @xmath0prolog @xcite ; without the latter one , a runtime check is needed to insure that every dynamically created goal is an acceptable one .
we can avoid putting ` : - ` inside arguments of predicates by writing the above term as .... ( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm t a b p ) ( eq t a b )
< < = = hastype a t , hastype b t , proves p ( eq t b a ) ) .... where ` < < = = ` is a new infix operator of type ` o - > o ` .
but this , in turn , means that the subgoal ` ( inference lemmaproof ) ` of the ` lemma_pf ` clause in program [ lemmafig ] will no longer check the lemma , since ` <
< = = ` has no operational meaning .
to handle such goals , we add the three constants declared at the beginning of program [ cinterp ] , which introduce both forward and backward implication arrows , and a new atomic predicate ` cl ` of type ` o - > o ` , and we introduce the two clauses that follow these declarations to interpret our new arrows as @xmath0prolog implication .
.... type = = > > o - > o - > o. infixr = = > >
type < < = = o - > o - > o. infixl < < = = 0 .
type cl o - > o. ( d = = > > g ) : - ( cl d ) = > g. ( g < < = = d ) : - ( cl d ) = > g. type backchain o - > o - > o. proves p a : - cl cl , backchain ( proves p a ) cl .
hastype x t : - cl cl , backchain ( hastype x t ) cl .
assump g : - cl cl , backchain ( assump g ) cl .
backchain g g. backchain g ( pi d ) : - backchain g ( d x ) .
( a , b ) : - backchain g a ; backchain g b. backchain g ( h < < = = g1 ) : - backchain g h , g1 .
backchain g ( g1 = = > >
h ) : - backchain g h , g1 .
.... [ cinterp ] note that although it would have been more direct , we did not add : .... ( d = = > > g ) : - d = > g. .... because of the teyjus restriction mentioned above that variables can not appear at the head of the left of an implication .
the use of the ` cl ` `` wrapper '' solves the problem created by this restriction , but requires us to implement an interpreter to handle clauses of the form ` ( cl a ) ` .
the remaining clauses in program [ cinterp ] implement this interpreter . since the type of ` ( inference proof ) ` is ` o ` , the term ` inference ` might conceivably contain subterms which are @xmath0prologclauses . of course , in teyjus these clauses will not contain ` : - ` or ` = > ` , but they may contain ` < < = = ` and ` = = > > ` , which get interpreted via the clauses of program [ cinterp ] .
they could also , for example , contain any other @xmath0prolog code including input / output operations . executing ` ( inference proof ) ` can not lead to unsoundness if the resulting proof checks , it is still valid .
but there are some contexts where we wish to restrict the kind of program that can occur inside a proof and be run when the proof is checked . for example , in a proof - carrying - code system , the code consumer might not want proof checking to cause @xmath0prolog to execute code that accesses private local resources . to limit the kind and amount of execution possible in the executable part of a lemma , we introduce the ` valid_clause ` predicate of type ` o - > o ` ( program [ validclause ] ) .
.... valid_clause ( pi c ) : - pi x\ valid_clause ( c x ) .
valid_clause ( a , b ) : - valid_clause a , valid_clause b. valid_clause ( a < < = = b ) : - valid_clause a , valid_clause b. valid_clause ( a = = > > b ) : - valid_clause a , valid_clause b. valid_clause ( proves q a ) .
valid_clause ( hastype x t ) .
valid_clause ( assump ( proves q a ) ) .
.... [ validclause ] a clause is valid if it contains ` pi ` , _ comma _ , `
< < = = ` , ` = = > > ` , ` proves ` , ` hastype ` , ` assump ` , and nothing else .
of course , a ` proves ` or ` assump ` clause contains subexpressions of type ` pf ` and ` tm ` , and a ` hastype ` clause has subexpressions of type ` tm ` and ` tp ` , so all the constants in proofs , terms , and types of our object logic are also permitted .
absent from this list are @xmath0prolog input / output ( such as ` print ` ) and the semicolon ( backtracking search ) .
the ` valid_clause ` restriction is the reason that we only need new clauses for the ` proves ` , ` hastype ` , and ` assump ` predicates in program [ cinterp ] .
we must add at least these three because they are used for checking nodes in a proof that require using the clauses added dynamically via the ` cl ` predicate . including no other predicates in the ` valid_clause ` definition guarantees that we need no other new clauses with ` cl ` subgoals .
because of the introduction of `
< < = = ` , ` = = > > ` , and ` valid_clause ` , we modify the clause in program [ lemmafig ] for checking lemmas . the new clause is shown in program [ teyjuslemmafig ] .
.... proves ( lemma_pf inference lemmaproof restproof ) c : - pi name\ ( valid_clause ( inference name ) ) , inference lemmaproof , pi name\ ( cl ( inference name ) = > ( proves ( restproof name ) c ) ) . ....
[ teyjuslemmafig ] the first subgoal is new ; it ` pi`-binds ` name ` and checks to see if the new lemma applied to ` name ` is valid .
the only other modification is in the last subgoal , which adds the lemma as a new clause via the ` cl ` predicate . since all lemmas will be added via ` cl ` , the only way to use them is via the ` proves ` clause in program [ cinterp ] . using that clause , the ` ( cl cl ) ` subgoal looks up the lemmas that have been added , one at a time , and tries them out via the ` backchain ` predicate .
this predicate processes the clauses in a manner similar to the @xmath0prolog language itself . in terzo , using this interpreter is less efficient than the direct implementation in program [ lemmafig ] . in teyjus , the interpreter is required , but when compiled , the code runs faster than either terzo version . in summary , our technique allows lemmas to be contained _ within _ the proof . we do not need to install new `` global '' lemmas into the proof checker .
the dynamic scoping also means that the lemmas of one proof can not interfere with the lemmas of another , even if they have the same names .
this machinery uses several interesting features of @xmath0prolog : [ [ polymorphism . ] ] polymorphism .
+ + + + + + + + + + + + + the type of the ` lemma_pf ` constructor uses polymorphism to indicate that proof constructors introduced for lemmas can have different types . [
[ meta - level - formulas - as - terms . ] ] meta - level formulas as terms .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + lemmas such as symmetry of equality occur inside proofs as an argument to the ` lemma_pf ` constructor in the following form . ....
( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm t a b p ) ( eq t a b ) < < = = hastype a t , hastype b t , proves p ( eq t b a ) ) .... it is just a data structure ( parameterized by ` symm ` ) ; it does not `` execute '' anything , in spite of the fact that it contains the @xmath0prolog quantifier ` pi ` and our new connective ` < < = = ` .
this gives us the freedom to write lemmas using syntax very similar to that used for writing primitive inference rules .
handling the new constants for `
< < = = ` and ` = = > > ` is easy enough operationally .
however , it is an inconvenience for the user , who must use different syntax in lemmas than in inference rules .
this inconvenience is avoided in terzo .
[ [ dynamically - constructed - goals . ] ] dynamically constructed goals .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + when the clause from program [ teyjuslemmafig ] for the ` lemma_pf ` proof constructor checks the proof of a lemma by executing the goal ` ( inference lemmaproof ) ` , we are executing a goal that is built from a run - time - constructed data structure . `
inference ` will be instantiated with terms such as the one above representing the symmetry lemma .
it is only when such a term is applied to its proof and thus appears in `` goal position '' that it becomes the current subgoal on the execution stack .
[ [ dynamically - constructed - clauses . ] ] dynamically constructed clauses .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + when , having successfully checked the proof of a lemma , the ` lemma_pf ` clause executes .... cl ( inference name ) = > ( proves ( restproof name ) c ) .... it is adding a dynamically constructed clause to the @xmath0prolog database .
although it is not the case for terzo or teyjus , if a metalanguage were to prohibit all terms having ` o ` in their types as arguments to a predicate , it would still be possible to implement lemmas using our approach .
appendix [ dynamic ] illustrates by showing an interpreter which extends program [ cinterp ] to handle this extra restriction .
new constants must be introduced not only for implication but also for every meta - level connective .
note that when meta - level formulas are not allowed , there is no possibility for dynamically created goals or clauses .
twelf for example , does not allow meta - level formulas as terms and is also not polymorphic , and thus the approach described in this section can not be used , but the approach of appendix [ dynamic ] could . instead , as we will see in section [ sec
: twelf ] , twelf provides alternative features which we can use to implement lemmas .
the ` symm ` proof constructor in theorem [ thm : symm ] is a bit unwieldy , since it requires ` t ` , ` a ` , and ` b ` as arguments .
we can imagine writing a primitive inference rule .... proves ( symm p ) ( eq t a b ) : - hastype a t , hastype b t , p proves ( eq t b a ) .
.... using the principle that the proof checker does nt need to be told ` t ` , ` a ` , and ` b ` inside the proof term , since they can be found in the formula to be checked .
then , in theorem [ thm : symm ] , ` ( symm intty j i q ) ` would be ` ( symm q ) ` . therefore we add three new proof constructors`elam ` , ` extract ` , and ` extractgoal`as shown in program [ elam ] .
.... type elam ( a - > pf ) - > pf .
type extract tm - > pf - > pf .
type extractgoal o - > pf - > pf .
proves ( elam q ) a : - proves ( q b ) a. proves ( extract a p ) a : - proves p a. proves ( extractgoal g p ) a : - valid_clause g , g , proves p a. .... [ elam ] these can be used in the following stereotyped way to extract components of the formula to be proved .
first bind variables with ` elam ` , then match the target formula with ` extract ` .
theorem [ thm : symma ] is a modification of theorem [ thm : symm ] that makes use of these constructors . .... proves ( lemma_pf ( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm p )
( eq t a b )
< < = = hastype a t , hastype b t , proves p ( eq t b a ) ) ( p\ elam t\ elam a\ elam b\ ( extract ( eq t a b ) ( congr t b a ( eq t a ) p refl ) ) ) ( symm\ ( forall_i i\ ( forall_i j\ ( imp_i q\ ( symm q ) ) ) ) ) ) ( forall intty i\ forall intty j\ ( eq intty i j imp eq intty j i ) ) .
.... [ thm : symma ] note that we could eliminate the ` hastype ` subgoals from our new version of the ` symm ` lemma because we know them to be redundant as long as ` ( eq t a b ) ` was already typechecked .
the reason for keeping them is that the second subgoal of the clause in program [ teyjuslemmafig ] would fail without them ; the proof checking of the lemma requires these ` hastype ` assumptions . in encoding our core logic ,
it was possible to eliminate all such redundant subgoals .
the fact that such a shortcut is not possible in lemmas causes a tradeoff ; by keeping such lemmas out of the tcb and putting them in proofs , we are forcing the proof checker to do more work .
there seems to be no easy way to avoid this redundant work , though some ad - hoc optimizations to proof checking might be possible .
the ` extractgoal ` proof constructor asks the checker to run @xmath0prolog code to help construct the proof .
its implementation uses ` valid_clause ` to restrict what kinds of @xmath0prolog code can be run .
note , however , that ` valid_clause ` does not always eliminate code that loops and so its current implementation can not guarantee termination .
a stricter ` valid_clause ` would be necessary to achieve this .
the ` extractgoal ` proof constructor was useful for handling assumptions in the sequent calculus version of our object logic @xcite ; for natural deduction , the same need does not arise in the implementation of our core logic , but ` extractgoal ` is useful for implementing more complex lemmas .
although we have not done so , it would be interesting to further explore the possibility of creating more compact proofs by leaving out information that can be computed easily via code given as arguments to ` extractgoal ` . as another example of the use of lemmas , we can of course use one lemma in the proof of another , as shown by theorem [ thm : symmtrans ] .
.... proves ( lemma_pf ( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm p ) ( eq t a b ) < < = = hastype a t , hastype b t , proves p ( eq t b a ) ) ( p\ elam t\ elam a\ elam b\ ( extract ( eq t a b ) ( congr t b a ( eq t a ) p refl ) ) ) ( symm\ ( lemma_pf ( trans\ pi t\ pi a\ pi b\ pi c\ pi q1\ pi q2\ proves ( trans c q1 q2 ) ( eq t a b ) < < = = hastype a t , hastype b t , hastype c t , proves q1 ( eq t a c ) , proves q2 ( eq t c b ) ) ( c\q1\q2\ elam a\ elam b\ elam t\ ( extract ( eq t a b ) ( congr t b c ( eq t a ) ( symm q2 ) q1 ) ) ) ( trans\ ( forall_i i\ forall_i j\ forall_i k\ ( imp_i q1\ ( imp_i q2\ ( trans j ( symm q1 ) q2 ) ) ) ) ) ) ) ) ( forall intty i\ forall intty j\ forall intty k\ ( eq intty j i imp eq intty j k imp eq intty i k ) ) ) . .... [ thm : symmtrans ] the proof of the ` trans ` lemma expressing transitivity of equality uses the ` symm ` lemma .
the ` symm ` lemma is naturally polymorphic : it can express the idea that @xmath10 just as well as @xmath11 .
theorem [ thm : compose ] illustrates part of a proof which contains two lemmas whose proofs use ` symm ` at different types . ....
( lemma_pf ( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm p ) ( eq t a b )
< < = = hastype a t , hastype b t , proves p ( eq t b a ) ) ( p\ elam t\ elam a\ elam b\ ( extract ( eq t a b ) ( congr t b a ( eq t a ) p refl ) ) ) ( symm\ ( lemma_pf ( poly1\ proves poly1 ( forall ( intty arrow intty ) f\ forall ( intty arrow intty ) g\ ( eq ( intty arrow intty ) f g ) imp ( eq ( intty arrow intty ) g f ) ) ) ( forall_i f\ ( forall_i g\ ( imp_i q\ ( symm q ) ) ) ) ( poly1\ ( lemma_pf ( poly2\ proves poly2 ( forall ( intty arrow intty ) f\ forall intty x\ ( eq intty ( app intty f x ) x ) imp ( eq intty x ( app intty f x ) ) ) ) ( forall_i f\ ( forall_i x\ ( imp_i q\ ( symm q ) ) ) ) ( poly2\ ... ) ) ) ) ) ) .... [ thm : compose ] in our previous work @xcite , because we represented object - level types as meta - level types , we were unable to allow polymorphism in lemmas at all . to do so
would have required a metalanguage with more general non - prenex polymorphism . to handle theorem [ thm : compose ] required two copies of the ` symm ` lemma , one at each type . in principle , we do not need lemmas at all .
instead , we can replace each subproof of the form ` ( lemma_pf i l r ) ` with the term ` ( r l ) ` , which replaces each use of a lemma with its proof .
this approach , however , adds undesirable complexity to proofs .
but , using this fact it should be straightforward to prove the correspondence between proofs with the ` lemma_pf ` constructor and proofs without , which would directly extend soundness and adequacy results to our system with lemmas .
definitions are another important mechanism for structuring proofs to increase clarity and reduce size .
if some property ( of a base - type object , or of a higher - order object such as a predicate ) can be expressed as a logical formula , then we allow the introduction of an abbreviation to stand for that formula .
we start by presenting a motivating example ( section [ defsexample ] ) , which leads us to our definition mechanism in @xmath0prolog ( section [ defsteyjus ] ) .
we also discuss two simpler versions of our definition mechanism ( sections [ defssimple ] and [ defsatomic ] ) , which allow us to have a smaller tcb , but which require more work to use .
we can express the fact that @xmath2 is an associative function by the formula @xmath12 this will only be a valid expression if @xmath2 has type @xmath13 . putting this formula in @xmath0prolog notation and expressing the type constraint on @xmath2 , we get the following provable @xmath0prolog typechecking goal . .... pi f\ pi t\ ( pi x\ pi y\ hastype
x t = > hastype y t = > hastype
( f x y ) t ) = > hastype ( forall t x\ forall t y\ forall t z\ eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) form . .... to make this into a definition , the first step is to associate some name , say ` assoc ` , with the definition body ( which is the first argument of the last ` hastype ` above ) .
we associate a name to a body of a definition in the same way we associated a new proof constructor with the proof it stood for . if we follow exactly the pattern of the ` symm ` lemma introduced at the beginning of section [ lemmas ] , we abstract out the body of the definition and obtain the following query . ....
( typeinf = ( assoc\ pi f\ pi t\ hastype ( assoc f t ) form
< < = = pi x\ pi y\ ( hastype x t = = > > hastype y t = = > > hastype ( f x y ) t ) ) , def = ( f\t\ ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) , query = ( typeinf def ) , query ) .... ` typeinf ` is the typechecking query above with ` = > ` replaced by ` = = > > ` or ` < < = = ` , and the abstraction ` assoc ` replacing the body of the definition . ` def ` contains the body abstracted with respect to the function ` f ` and type ` t ` and ` ( typeinf def ) ` is exactly the typechecking subgoal above ( except for the use of ` = = > > ` and ` < < = = ` ) . if all we wanted was a typechecking lemma to typecheck expressions of the form given by ` def ` , then we could use our lemma mechanism directly . ....
( lemma_pf ( assoc\ pi f\ pi t\ hastype ( assoc f t ) form < < = = pi x\ pi y\ ( hastype x t = = > > hastype y t = = > > hastype ( f x y ) t ) ) ( f\t\ ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) ( assoc\ ... .... this example shows that we can have typechecking lemmas in addition to proof checking lemmas .
it also motivates our definition mechanism shown next , which we obtain by adding the ability to replace a name with the expression it represents and vice versa .
we introduce a new proof constructor ` def_pf ` and a new proof term ` def ` to represent equality between a name and its definition .
this definition mechanism is implemented by the clauses in program [ definition - fig ] .
.... type def_pf tp - > ( a - > o ) - > a - > ( a - > pf ) - > pf .
type def pf .
type def_to_eqclause tp - > a - > a - > o - > o. def_to_eqclause t dname def ( pi clause ) : - pi x\ ( def_to_eqclause t ( dname x ) ( def x ) ( clause x ) ) .
def_to_eqclause t dname def ( proves def ( eq t dname def ) ) .
proves ( def_pf t typeinf term restproof ) c : - pi name\ ( valid_clause ( typeinf name ) , typeinf term , def_to_eqclause t name term ( eqclause name ) , cl ( typeinf name ) = > cl ( eqclause name ) = > ( proves ( restproof name ) c ) ) .
.... [ definition - fig ] the arguments to ` def_pf ` are similar to the arguments to ` lemma_pf ` , but also include one more for the type of the body of the definition ( after it is applied to all its arguments ) . in the clause for proof checking ` def_pf ` nodes , the first two subgoals are similar to ` lemma_pf ` nodes . here
, they check that the typechecking clause is valid and that ` term ` ( the body of the definition ) is correctly typed .
the third clause computes the clause for expressing definitional equality using the ` def_to_eqclause ` program .
the fourth subgoal for proof checking definitions adds both the typechecking clause and the equality clause before checking the rest of the proof . like ml , @xmath0prolog has parametric polymorphism ( in the syntactic sense ) .
but unlike ml , @xmath0prolog does not have the parametricity property .
a polymorphic function can examine the structure of its argument .
we illustrate with a simple example : a function that tells the arity ( number of function arguments ) of an arbitrary value .
.... type arity a - > int - > o. arity f n : - arity ( f x ) n1 , n is n1 + 1 .
arity x 0 . .... the first clause can only be used when ` f ` is a function ; the second clause matches any value .
the ` def_to_eqclause ` clauses uses this exact feature of @xmath0prolog s polymorphism .
it first uses the meta - level type of ` def ` to apply ` def ` to as many arguments as possible .
the first clause introduces new variables to serve as these arguments . once it is applied to all of its arguments , the second clause forms the equality clause using the type , the name , and the body of the definition . for our example , the computed clause is .... eqclause = ( assoc\ ( pi f\ pi t\ proves def ( eq form ( assoc f t ) ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) ) ) . .... to ensure that there is only one solution to the ` arity ` predicate above and likewise the ` def_to_eqclause ` predicate in program [ definition - fig ] , we could have used the logic programming cut ( ` ! ` ) operator at the end of the first clause for each predicate .
we have omitted it here because ` def_to_eqclause ` is only be used in our proof checker , which is written to avoid the need for backtracking . to use definitions in proofs
we introduce two new lemmas : ` def_i ` to replace a formula with the definition that stands for it ( or viewed in terms of backward proof , to replace a defined name with the term it stands for ) , and ` def_e ` to expand a definition in the forward direction during proof construction .
their proofs are shown in program [ defs - fig ] . ....
( lemma_pf ( def_i\ pi t\ pi name\ pi b\ pi p\ pi q1\ pi q2\ proves ( def_i t name b p q1 q2 ) ( p name )
< < = = proves q1 ( eq t name b ) , hastype name t , hastype b t , proves q2 ( p b ) ) ( t\name\b\p\q1\q2\ ( congr t name b p q1 q2 ) ) ( def_i\ ( lemma_pf ( def_e\ pi t\ pi name\ pi b\ pi p\ pi q1\ pi q2\ proves ( def_e t name b p q1 q2 ) ( p b ) <
< = = proves q1 ( eq t name b ) , hastype name t , hastype b t , proves q2 ( p name ) ) ( t\name\b\p\q1\q2\ ( congr t b name p ( congr t name b ( eq t b ) q1 refl ) q2 ) ) ( def_e\ ... .... [ defs - fig ] theorem [ thm : def ] shows a proof using definitions . in this proof , ` f ` is a function symbol and ` t ` is a type , and the theorem is represented as a @xmath0prolog subgoal with a top - level implication , where the right hand side is a ` proves ` subgoal and the left hand side specifies the typing information about ` f ` which must hold in order for the proof in the ` proves ` subgoal to be valid . .... pi f\ pi t\ ( pi x\ pi y\ hastype
x t = > hastype y t = > hastype ( f x y ) t ) = > ( proves ( lemma_pf ... symm\ ( lemma_pf ... trans\ ( lemma_pf ... def_i\ ( lemma_pf ...
def_e\ ( def_pf form ( assoc\ pi f\ pi t\ hastype ( assoc f t ) form < < = = pi x\ pi y\ ( hastype x t = = > > hastype y t = = > > hastype
( f x y ) t ) ) ( f\t\ ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) ( assoc\ ( lemma_pf ( assoc_inst\ pi f\ pi t\ pi a\ pi b\ pi c\ pi q\ proves ( assoc_inst f q ) ( eq t ( f a ( f b c ) ) ( f ( f a b ) c ) ) < < = = hastype a t , hastype b t , hastype c t , pi x\ pi y\ ( hastype x t = = > > hastype y t = = > > hastype ( f x y ) t ) , proves q ( assoc f t ) ) ( f\q\ ( elam t\ elam a\ elam b\ elam c\ ( extract ( eq t ( f a ( f b c ) ) ( f ( f a b ) c ) ) ( forall_e t ( z\ ( eq t
( f a ( f b z ) ) ( f ( f a b ) z ) ) ) ( forall_e t ( y\ ( forall t z\ ( eq t ( f a ( f y z ) ) ( f ( f a y ) z ) ) ) ) ( forall_e t ( x\ ( forall t y\ ( forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) ) ( def_e form ( assoc f t ) ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ( x\x ) def
q ) a ) b ) c ) ) ) ) ( assoc_inst\ ( imp_i q1\ ( forall_i a\ ( imp_e ( assoc f t ) ( imp_i q2\ ( trans ( f ( f a a ) ( f a a ) ) ( assoc_inst f q2 ) ( assoc_inst f q2 ) ) ) ( def_i form ( assoc f t ) ( forall t a\ forall t b\ forall t c\ ( eq t ( f a ( f b c ) ) ( f ( f a b ) c ) ) ) ( x\x ) def q1 ) ) ) ) ) ) ) ) ) ) ) ) ( ( forall t a\ forall t b\ forall t c\ eq t ( f a ( f b c ) ) ( f ( f a b ) c ) ) imp ( forall t a\ eq t
( f a ( f a ( f a a ) ) ) ( f ( f ( f a a ) a ) a ) ) ) ) .... [ thm : def ] the proof ( the first argument to the ` proves ` predicate ) contains a series of four lemmas which we have already seen , followed by the definition of associativity , followed by a fifth lemma about associativity ( ` assoc_inst ` ) , followed by the main body of the proof .
the ` def_i ` lemma is used in the main body of the proof . in general ,
proof checking using the ` def_i ` lemma means that the proof being checked must match the term ` ( def_i t name b p q1 q2 ) ` , which is the first argument ( the proof term ) of the head of the ` proves ` clause implementing the ` def_i ` lemma in program [ defs - fig ] .
this match determines the terms matching ` p ` and ` name ` .
the formula being proved must be a formula that matches the term ` ( p name ) ` , which is the second argument of the head of the ` proves ` clause implementing the ` def_i ` lemma in program [ defs - fig ] . here
` name ` is not always simply a variable name , but is actually the definition name applied to all of its arguments to form a term of type ` tm ` . in our example , ` assoc ` has type .... ( tm - > tm - > tm ) - > tp - > tm . .... at the point that proof checking of the body of the proof uses the ` def_i ` lemma
, the formula to be checked is ` ( assoc f t ) ` .
the term that corresponds to ` ( p name ) ` in this example is ` ( x\x)(assoc f t ) ` , which matches this formula .
proof checking proceeds by finding a proof of the goal of the form .... ( proves q1 ( eq form ( assoc f t ) b ) ) .... which is proved simply by matching with the @xmath0prolog equality assumption added when the ` assoc ` definition was processed by the ` proves ` clause for ` def_pf ` .
next , the two typechecking subgoals of the ` def_i ` clause are solved . solving the first , ` ( hastype ( assoc f t ) form ) ` , requires using the @xmath0prolog type inference assumption which was also added when the ` assoc ` definition was processed by the ` proves ` clause for ` def_pf ` .
finally , the rest of the proof , is checked via the subgoal of the form ` ( proves q2 ( p b ) ) ` , where the formula to be checked has the definition name replaced by its body .
the ` def_e ` lemma is used in the proof of the ` assoc_inst ` lemma .
its use in proof checking is similar to ` def_i ` .
the main difference is that the formula to be checked must match the term ` ( p b ) ` , i.e. , the formula contains an instance or instances of the body of the definition , and in the subgoal to be checked , the body of the definition is replaced with the name of the definition . as another example of definitions , program [ def : and ] shows the definition of logical conjunction for the object logic using the ` def_pf ` proof constructor . ....
( def_pf form ( and\ pi a\ pi b\ hastype ( and a b ) form
< < = = hastype a form , hastype b form ) ( a\b\ ( forall form c\ ( ( a imp b imp c ) imp c ) ) ) ( and\ ... .... [ def : and ] other connectives such as disjunction , negation , and existential quantification can also be defined , and the rules for introduction and elimination of these connectives can be proved as lemmas . the new primitives and clauses in program [ definition - fig ] provide a convenient way of incorporating definitions , but actually are not needed at all . instead , for each new definition , it is possible to introduce a special lemma to handle that definition .
these special lemmas are quite complex and we do not want to require the user to come up with them
. for illustration , theorem [ thm : altdef ] shows the part of the proof that replaces the ` def_pf ` node in theorem [ thm : def ] . .... ... ( lemma_pf ... def_e\ ( lemma_pf ( define_assoc\ pi q\ pi b\ proves ( define_assoc q ) b < < = = pi d\ pi q\ ( pi f\ pi t\ ( proves q ( eq form ( d f t ) ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) ) ) = = > > ( pi f\ pi t\ hastype ( d f t ) form
< < = = pi x\ pi y\ hastype x t = = > > hastype y
t = = > > hastype ( f x y ) t ) = = > > proves ( q d q ) b ) ( q\ ( q ( f\t\ ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) refl ) ) ( define_assoc\ ( define_assoc ( assoc\q\ ( lemma_pf ( assoc_inst\ ... .... [ thm : altdef ] this part of the proof includes the specialized lemma , called ` define_assoc ` , and shows that it is used immediately after it is defined .
the bound variable ` assoc ` represents the name for the new definition , and the bound variable ` q ` represents a proof of equality between the definition name and its body .
the new proof contains no use of the ` def_pf ` or ` def ` proof constructors .
occurrences of ` def ` in theorem [ thm : def ] are replaced with ` q ` .
this change , although not shown in theorem [ thm : altdef ] , is the only other change required to obtain the complete alternate proof .
we omit a detailed explanation of the ` define_assoc ` lemma and simply note that it is fairly complex and increases the size of this example proof .
also , this lemma is similar in structure to the simpler ` define ` lemma described below in section [ defsatomic ] additional programming can make this alternative way of incorporating definitions easier to use .
in particular , it is possible to write a program to transform proofs that use ` def_pf ` and ` def ` to proofs that use only specialized lemmas such as the one in theorem [ thm : altdef ] .
such a program would allow us to remove program [ definition - fig ] from the tcb .
for the special class of defined terms that have meta - level type ` tm ` , which we call _ atomic _ definitions , it is easy to eliminate the need for ` def_pf ` and ` def ` because it is possible to include one new general lemma that replaces them .
for example , we can express associativity of integers as the following term .... lam f\ forall intty x\ forall intty y\ forall intty z\ eq intty ( app intty ( app intty f x ) ( app intty ( app intty f y ) z ) ) ( app intty ( app f intty ( app intty ( app intty f x ) y ) ) z ) ) ) .... where ` f ` has meta - type ` tm ` and object type ` ( intty arrow intty arrow intty ) ` , and the ` app ` constructor is used to apply ` f ` to its arguments . if we specialize theorem [ thm : def ] to integers , theorem [ thm : newdef ] shows the part of the proof of this new theorem that replaces what is shown in theorem [ thm : altdef ] . .... ... ( lemma_pf ... def_e\ ( lemma_pf ( define\ pi t\ pi f\ pi q\ pi b\ proves ( define t f q ) b < < = = hastype f t , pi d\ pi q\
( hastype d t = = > > proves q ( eq t d f ) = = > > proves ( q d q ) b ) ) ( t\f\p\ ( p f refl ) ) ( define\ ( define ( ( intty arrow intty arrow intty ) arrow form ) ( lam f\ forall intty x\ forall intty y\ forall intty
z\ eq intty ( app intty ( app intty f x ) ( app intty ( app intty f y ) z ) ) ( app intty ( app intty f ( app intty ( app intty f x ) y ) ) z ) ) ) ( assoc\q\ ( lemma_pf ( assoc_inst\ ... .... [ thm : newdef ] the parts of the proof not shown are similar to theorems [ thm : def ] and [ thm : altdef ] , but modified to use the new type of the bound variable ` f ` , which has the same type as the bound ` f ` in the definition . in general , to check a proof using the ` define ` lemma , which has the following form .... ( define t term ( name\ eqproof\ ( restproof name eqproof ) ) ) .... the system interprets the `` ` pi d ` '' within the ` define ` lemma to create a new atom ` d ` to stand for the ` name ` .
the new atom ` q ` is also introduced to stand for a proof that the name is equal to the body of the definition , and ` ( proves q ( eq t d term ) ) ` is added to the clause database .
finally , @xmath4-conversion substitutes ` d ` for ` name ` and ` q ` for ` eqproof ` within ` restproof ` and the resulting proof is checked . in proof checking the new proof , instead of subproofs of the form .... ( proves def ( eq form ( assoc f t ) b ) )
.... that would be generated by proof checking thereom [ thm : def ] , or subproofs of the form .... ( proves q ( eq form ( assoc f t ) b ) ) .... that would be generated by proof checking thereom [ thm : altdef ] , in theorem [ thm : newdef ] we have subproofs of the form .... ( proves q ( eq ( ( intty arrow intty arrow intty ) arrow form ) assoc b ) ) .... where ` q ` here is the name of the proof term introduced inside the ` define ` proof node . in general , having a single ` define ` lemma that can be used by all atomic definitions is simpler , but the atomic forms of definitions are larger and harder to read . in the case of ` assoc ` , the atomic version is three lines , while the original version is one line long . in our previous work
@xcite , having to choose between the version of ` assoc ` that used ` app ` and the one that did nt was not an issue , since there were no ` app ` and ` lam ` constructors . instead application and abstraction were encoded directly using application and abstraction at the meta - level .
also , there was no reason to include a separate ` def_pf ` proof constructor ; the ` define ` lemma was sufficient for introducing all definitions .
although this allowed a simpler version of definitions , we were unable to allow polymorphism in definitions , which is desirable in definitions for the same reason it is desirable in lemmas .
our previous encoding also did not allow definitions for object - level types .
for example , in the domain of proof - carrying code , we have declarations like this one .... hastype has_mltype ( ( exp arrow form ) arrow ( exp arrow exp ) arrow exp arrow ( ( exp arrow form ) arrow ( exp arrow exp ) arrow exp arrow form ) arrow form .
.... types like this arise because we encode types of the programming language we are reasoning about ( in this case ml ) as predicates which themselves take predicates as arguments . in our new version
, it is possible to handle definitions at meta - type ` tp ` ; we would need a new proof constructor and a new proof checking clause similar to the one for the ` def_pf ` proof constructor in program [ definition - fig ] .
adding type definitions would also require adding reasoning about equality of types into our typechecking clauses .
the lemma and definition mechanisms provide ways to store lemmas and definitions inside proofs .
packaging proofs in this way makes it straightforward to communicate proofs , and keeps the proof checking machinery ( the tcb ) simple , which is important for our proof - carrying code application .
thus far , all the @xmath0prolog code in programs [ core - types ] , [ core - rules ] , [ lemmafig ] , [ cinterp ] , [ validclause ] , [ teyjuslemmafig ] , [ elam ] , and [ definition - fig ] is inside the tcb . a good environment for building proofs is also essential , and this part of the code can be outside the tcb .
we do nt have to be as careful because we know that any proofs we build in our theorem proving environment have to be checkable by the proof checking code presented so far . as we build a library of lemmas and definitions , we clearly do nt want to store every lemma and definition inside every proof that uses them .
instead , for lemmas that have general applicability like ` symm ` , we would like to store them each once and allow them to be used in other proofs as needed . to do so
, we provide predicates for stating each definition and lemma . to use these predicates ,
we must introduce new constants for definition and lemma names .
program [ prog : store ] contains the declarations of these new predicates , and two examples which use them . ....
type def_lemma a - > ( a - > o ) - > a - > o. type def_definition tp - > a - > ( a - > o ) - > a - > o. type symm pf - > pf .
def_lemma symm ( symm\ pi t\ pi a\ pi b\ pi p\ proves ( symm p ) ( eq t a b )
< < = = hastype a t , hastype b t , proves p ( eq t b a ) ) ( p\ elam a\ elam b\ elam t\ ( extract ( eq t a b ) ( congr t b a ( eq t a ) p refl ) ) ) .
type assoc ( tm - > tm - > tm ) - > tp - > tm .
def_definition form assoc ( assoc\ pi f\ pi t\ hastype ( assoc f t ) form < < = = pi x\ pi y\ ( hastype x t = = > > hastype y t = = > > hastype ( f x y ) t ) ) ( f\t\ ( forall t x\ forall t y\ forall t z\ ( eq t ( f x ( f y z ) ) ( f ( f x y ) z ) ) ) ) . .... [ prog : store ] @xmath0prolog s polymorphism is used in these predicates . the first argument to ` def_lemma ` is the lemma name , and the next two arguments correspond to the ` inference ` and ` lemmaproof ` arguments to the ` lemma_pf ` constructor .
the arguments to ` def_definition ` are the definition name ( the second argument ) and arguments that correspond to the first three arguments of the ` def_pf ` constructor ( arguments 1 , 3 , and 4 here )
. then we can write programs to manipulate these lemmas and definitions in various ways .
for example , if we want to package a proof as a single term with all the definitions and lemmas it depends on inside it , we must write a program to do so .
the resulting proof should not contain any constants like ` symm ` and ` assoc ` ; instead lemma and definition names must be bound variables inside occurrences of the ` lemma_pf ` and ` def_pf ` proof constructors .
we do not present the `` packaging '' program here , but instead present a simpler program that illustrates some of the programming techniques required for manipulating lemmas and definitions stored in this way .
program [ prog : check ] contains a program for checking a proof .
it does nt check the lemmas that the proof depends on , but could be easily modified to do so .
.... type done_def a - > o. type done_lemma a - > o. type check_lem a - > o. type check_lem_aux b - > a - > ( a - > o ) - > a - > o. check_lem name : - def_definition t dname inference def , not ( done_def dname ) , ! , def_to_eqclause t dname def eqclause , done_def dname = > cl ( inference dname ) = > cl eqclause = > check_lem name .
check_lem name : - def_lemma lname inference lemmaproof , check_lem_aux name lname inference lemmaproof .
check_lem_aux name name inference lemmaproof : - ! , pi name\ ( valid_clause ( inference name ) ) , ( inference lemmaproof ) .
check_lem_aux name lname inference lemmaproof : - ! , not ( done_lemma lname ) , !
, done_lemma lname = > cl ( inference lname ) = > check_lem name . .... [ prog : check ] the trick of using prolog cut ( ` ! ` ) along with the predicates ` done_def ` and ` done_lemma ` allows us to process a list of clauses in the order they appear in the database . the first clause for ` check_lem ` looks for the next definition and each time it finds a new one , it adds the corresponding typechecking clause and equality clause . the second ` check_lem ` clause is used once all definitions have been added .
it finds the next lemma and uses ` check_lem_aux ` to see if the next lemma is the one that should be checked .
if so , the proof is checked ; if not , the proof checking clause for the lemma is added to the database and ` check_lem ` is called to process the next lemma .
the _ logical framework _
( lf ) @xcite is another example of a metalanguage in which it is possible to encode a wide variety of logics .
the twelf system @xcite is an implementation of lf which provides logic programming capabilities , many of which are similar to @xmath0prolog . in this section ,
we compare the encoding of our core logic in @xmath0prolog to a corresponding encoding in twelf , discuss lemmas and definitions in twelf , and compare the programming environments of these two languages .
lf is a @xmath0-calculus with dependent types .
a dependent type in lf has the structure @xmath14 where @xmath15 and @xmath16 are types and @xmath17 is a variable of type @xmath15 bound in this expression .
the type @xmath16 may contain occurrences of @xmath17 .
this structure represents a `` functional type . ''
if @xmath2 is a function of this type , and @xmath18 is a term of type @xmath15 , then @xmath19 ( @xmath2 applied to @xmath18 ) has the type @xmath20b$ ] , which represents the type @xmath16 where all occurrences of @xmath17 are replaced by @xmath18 .
thus the argument type is @xmath15 and the result type _ depends _ on the value input to the function . if @xmath17 does nt occur in @xmath16 , this type is often abbreviated using the usual type arrow : @xmath21 .
the extra expressiveness of dependent types allows object - level types to be expressed more directly , eliminating the need for any typechecking clauses like the ` hastype ` clauses of program [ core - rules ] .
the twelf constructor declarations in program [ core - twelf ] illustrate the use of dependent types for encoding our object logic .
.... tp : type .
tm : tp - > type .
form : tp .
pf : tm form - > type .
intty : tp .
arrow : tp - > tp - > tp .
% infix right 14 arrow .
pair : tp - > tp - > tp .
eq : tm t - > tm t - > tm form .
imp : tm form - > tm form - > tm form .
% infix right 10 imp .
forall : ( tm t - > tm form ) - > tm form .
false : tm form .
lam : ( tm t1 - > tm t2 ) - > tm ( t1 arrow t2 ) .
app : tm ( t1 arrow t2 ) - > tm t1 - > tm t2 .
mkpair : tm t1 - > tm t2 - > tm ( pair t1 t2 ) .
fst : tm ( pair t1 t2 ) - > tm t1 .
snd : tm ( pair t1 t2 ) - > tm t2 .
refl : pf ( eq x x ) .
beta : pf ( eq ( app ( lam f ) x ) ( f x ) ) .
fstpair : pf ( eq ( fst ( mkpair x y ) ) x ) .
sndpair : pf ( eq ( snd ( mkpair x y ) ) y ) .
surjpair : pf ( eq ( mkpair ( fst z ) ( snd z ) ) z ) . congr : { h : tm t - > tm form } pf ( eq x z ) - > pf ( h z ) - > pf ( h x ) . imp_i : ( pf a - > pf b ) - > pf ( a imp b ) . imp_e : pf ( a imp b ) - > pf a - > pf b. forall_i : ( { y : tm t}pf ( a y ) ) - > pf ( forall a ) .
forall_e : pf ( forall a ) - > { y : tm t}pf ( a y ) . ....
[ core - twelf ] felty and miller @xcite show how to transform an lf object logic into an encoding in a higher - order logic which is a sublogic of the one implemented by @xmath0prolog .
the discussion in this section is informal , but in appendix [ app : compare ] , we use this transformation to provide a formal basis for comparing our two encodings . although typechecking clauses are not needed here , the proof checking operation is more complicated in twelf since it requires type reconstruction for dependent types .
twelf has its own built - in definition mechanism , which can be used for both lemmas and definitions in the object logic .
program [ twelf - def ] contains a twelf version of the definition of ` assoc ` and the ` symm ` lemma .
.... % abbrev assoc : ( tm t - > tm t - > tm t ) - > tm form = [ f:(tm t - > tm t - > tm t ) ] ( forall [ a : tm t ] forall [ b : tm t ] forall [ c : tm t ] ( eq ( f a ( f b c ) ) ( f ( f a b )
c ) ) ) .
symm : pf ( eq x y ) - > pf ( eq y x ) = [ q : pf ( eq x y ) ] ( congr ( [ z : tm t ] ( eq y z ) ) q refl ) . .... [ twelf - def ] the ` abbrev ` directive is required in some definitions for technical reasons , which we do not describe here .
there are three parts to a definition : a constant naming the definition , its type , and its body ( an lf term ) .
a lemma is similar and contains its name , the formula representing the statement of the lemma ( which is a type in lf ) , and the proof ( an lf term ) . in twelf ,
a proof is simply a series of declarations and definitions , where the last one is the statement and proof of the main theorem .
this proof possibly depends on the lemmas and definitions that come before it .
each definition in the sequence has the form mentioned above : a name , a type , and the term which the name abbreviates when it appears in subsequent declarations .
the declarations defining the logical constants and primitive inference rules shown in program [ core - twelf ] ( which each have a type but no defining term ) are at the beginning of the sequence . in twelf , we can not package up a lemma and its proof , or a definition and its body , along with the rest of the proof , in the same way we did in @xmath0prolog .
the reason for this is that we can not introduce a ` lemma_pf ` or ` def_pf ` constructor because they require polymorphism at the meta - level , which twelf does not have . in our @xmath0prolog version
, we discussed naming each lemma and definition , including one copy of each in a library , and using it whenever needed .
we then presented a program which was able to check the proof of a theorem , assuming that lemmas and definitions were organized in this way . in twelf
, we do nt need a special program for checking proofs of lemmas .
one of the central meta - operations of twelf is to read in a series of declarations and definitions , and check each one as it is encountered .
proofs are fully checked by this operation . in twelf ,
other kinds of operations on proofs are limited .
many proof transformations that we can implement in @xmath0prolog are not programmable in twelf either because they require polymorphism or because they require manipulation of meta - level formulas .
manipulation of meta - level formulas is not possible in twelf because it requires quantification over such formulas ( i.e. , quantification over types containing ` type ` ) , which is not allowed .
although we have focussed on the lemma and definition mechanisms in @xmath0prolog and twelf , other aspects of the metalanguage are also relevant to our needs for proof generation and checking . for our application , proof - carrying code ,
we wish to prove theorems about machine instructions that add , subtract , and multiply ; and about load / store instructions that add offsets to registers .
therefore we require some rudimentary integer arithmetic in our logic .
some logical frameworks have powerful arithmetic primitives , such as the ability to solve linear programs @xcite or to handle general arithmetic constraints @xcite .
for example , twelf provides a complete theory of the rationals , implemented using linear programming @xcite . on the one hand ,
linear programming is a powerful and general proof technique , although it can increase the complexity of the tcb . on the other hand ,
synthesizing arithmetic from scratch is not easy .
we have also experimented with arithmetic in @xmath0prolog where we use the ` is ` predicate to provide some automatic simplifications .
parameterizable data structures with higher - order unification modulo @xmath4-equivalence provide an expressive way of representing formulas , predicates , and proofs .
we make heavy use of higher - order data structures with both direct sharing and sharing modulo @xmath4-reduction
. the implementation of the metalanguage must preserve this sharing ; otherwise our proof terms will blow up in size .
any logic programming system is likely to implement sharing of terms obtained by copying multiple pointers to the same subterm . in terzo , this can be seen as the implementation of a reduction algorithm described by wadsworth @xcite .
but we require even more sharing .
the similar terms obtained by applying a @xmath0-term to different arguments should retain as much sharing as possible .
therefore some intelligent implementation of higher - order terms within the metalanguage such as teyjus s use of explicit substitutions @xcite seems essential . in this paper , we have concentrated on an encoding of the logic used for proof checking , and discussed some operations on proofs .
but of course , we will also need to construct proofs .
for the proof - carrying code application , we need an automatic theorem prover to prove the safety of programs . for implementing this prover
, we have found that the prolog - style control primitives ( such as the cut ( ` ! ` ) operator and the ` is ` predicate ) , which are also available in @xmath0prolog , are quite important .
@xmath0prolog also provides an environment for implementing tactic - style interactive provers @xcite .
this kind of prover is useful for proving the lemmas that are used by the automatic prover .
twelf does not have many control primitives ; in fact , implementation of control primitives does not fit well into the twelf system design .
we have begun to experiment with an operator in twelf similar to prolog cut , to see if it will allow us to implement the automatic prover in the same way as in @xmath0prolog .
there is also no support for building interactive provers in twelf , so proofs of lemmas used by the automatic prover must be constructed by hand .
the logical frameworks discussed in this paper are promising vehicles for proof - carrying code , or in general where it is desired to keep the proof checker as small and simple as possible .
we have proposed a representation for lemmas and definitions that should help keep proofs small and well - structured , and each of these frameworks has features that are useful in implementing , or implementing efficiently , our machinery .
we have found the conciseness of the encoding in twelf to be particularly convenient , and because of that , we have used twelf for extensive proof development in our proof - carrying code application . as
programming with proofs becomes more important in the next phases of our system , @xmath0prolog will have more advantages .
we are currently investigating ways to combine the use of the two metalanguages .
the translation discussed in appendix [ app : compare ] will serve as the foundation for this combination .
to write a full interpreter , we extend program [ cinterp ] in section [ lemmasteyjus ] by introducing a new type ` goal ` and connectives which build terms of this type . in particular , we now give ` < < = = ` and ` = = > > ` the type ` goal - > goal - > goal ` . we also introduce a new constant ` ^^ ` for conjunction having the same type as the implication constructors .
finally , we introduce ` all ` for universal quantification having type ` ( a - > goal ) - > goal ` . in addition
, we change the type of ` backchain ` to ` goal - > goal - > o ` , and modify the clauses for the _ comma _ and ` pi ` to use the new constants . in the ` backchain ` clauses for `
< < = = ` and ` = = > > ` in program [ cinterp ] , the goal ` g1 ` which appears as an argument inside the head of the clause also appears as a goal in the body of the clause . in the full interpreter
, we can not do this . `
g1 ` no longer has type ` o ` ; it has type ` goal ` and is constructed using the new connectives . instead , we replace ` g1 ` with ` ( solveg g1 ) ` and implement the ` solveg ` predicate to handle the solving of goals . the new code for ` solveg ` and the modified code for ` backchain ` is in program [ interp ]
. .... kind goal type .
type = = > > goal - > goal - > goal .
infixr = = > >
type < < = = goal - > goal - > goal .
infixl < < = = 0 .
type ^^ goal - > goal - > goal .
infixl ^^ 3 .
type all ( a - > goal ) - > goal .
type cl goal - > o. type backchain goal - > goal - > o. type solveg goal - > o. type proves pf - > form - > goal .
type assume form - > goal .
type valid_clause goal - > goal .
solveg ( all g ) : - pi x\ ( solveg ( g x ) ) .
solveg ( g1 ^^ g2 ) : - solveg g1 , solveg g2 .
solveg ( d = = > > g ) : - ( cl d ) = > solveg g. solveg ( g < < = = d ) : - ( cl d ) = > solveg g. solveg g : - cl d , backchain g d. backchain g g. backchain g ( all d ) : - backchain g ( d x ) . backchain g ( a ^^ b ) : - backchain g a ; backchain g b. backchain g ( h < < = = g1 ) : - backchain g h , solveg g1 .
backchain g ( g1 = = > >
h ) : - backchain g h , solveg g1 . .... [ interp ] in order to use this interpreter to solve goals of the form ` ( proves p a ) ` , the ` proves ` predicate must be a constructor for terms of type ` goal ` , and the meta - level goal presented to @xmath0prolog must have the form ` ( solveg ( proves p a ) ) ` . similarly , inference rules must also be represented as objects of type ` goal ` and wrapped inside ` cl ` to form @xmath0prolog clauses .
several examples of clauses for inference rules are given in program [ inf - interp ] to illustrate . .... cl ( proves q a < < = = assump ( proves q a ) ) .
cl ( proves ( imp_i q ) ( a imp b ) < < = = all p\ ( assump ( proves p a ) = = > > proves ( q p ) b ) ) .
cl ( proves ( forall_i q ) ( forall t a ) < < = = all y\ ( hastype y t = = > > proves ( q y ) ( a y ) ) ) .
cl ( proves ( lemma_pf inference lemmaproof restproof ) c <
< = = all name\ ( valid_clause ( inference name ) ^^ inference lemmaproof ^^ ( inference name ) = = > > ( proves ( restproof name ) c ) ) ) . .... [ inf - interp ] the last clause is the new clause for handling lemmas in this setting . note that in this version , ` valid_clause ` constructs objects of type ` goal ` ; thus all the clauses for ` valid_clause ` must also be wrapped in ` cl ` .
as stated , the transformation in felty and miller @xcite can provide a formal basis for comparing our two encodings . in order to perform this transformation , we must consider a `` full '' lf encoding , which does not take advantage of the abbreviations that twelf allows .
just as the full lf encoding can be improved by using twelf s abbreviations , the @xmath0prolog program that results from the transformation can be improved by making several optimizations .
we discuss how the encoding presented in programs [ core - types ] and [ core - rules ] can be viewed as the application of the transformation , followed by performing several such optimizations . in both @xmath0prolog and twelf ,
all tokens in a clause or declaration beginning with uppercase letters are implicitly bound by universal quantifiers at the outermost level . in twelf , this implicit quantification is important for providing an encoding of the object logic that is readable and usable . to see why , consider the ` surjpair ` rule , which uses the ` mkpair ` , ` fst ` , and ` snd ` constants .
we can make the outermost quantification explicit in twelf , resulting in the declarations : .... mkpair : { t1:tp}{t2:tp}tm t1 - > tm t2 - > tm ( pair t1 t2 ) .
fst : { t1:tp}{t2:tp}tm ( pair t1 t2 ) - > tm t1 .
snd : { t1:tp}{t2:tp}tm ( pair t1 t2 ) - > tm t2 .
surjpair : { t1:tp}{t2:tp}{z : tm ( pair t1 t2 ) } pf ( eq ( pair t1 t2 ) ( mkpair t1 t2 ( fst t1 t2 z ) ( snd t1 t2 z ) ) z ) . .... this version of ` surjpair ` is quite a bit bigger than the one in program [ core - twelf ] .
explicitly including ` t1 ` and ` t2 ` means that ` mkpair ` , ` fst ` , and ` snd ` each take two extra type arguments , while ` surjpair ` takes three .
terms containing these constants must then take extra arguments which in this example causes redundancy in the type of ` surjpair ` because the same types appear many times .
implicit quantifiers make the encoding easier to read and work with .
in fact , in the version we used in our experiments , the fact that ` app ` could be represented as a binary constructor without loss of information allowed us to replace the ` app ` constant with an infix symbol , resulting in encoded terms that were syntactically even closer to the terms they represented .
we can not make ` app ` in the @xmath0prolog encoding infix because it takes three arguments .
( we discuss why it must take three arguments below . )
the explicit quantifiers that we have left out in program [ core - twelf ] are those that twelf can easily reconstruct . because of this reconstruction , however , a twelf typechecker ( proof checker ) has to work harder than it would if we used an explicit version .
these encodings illustrate a tradeoff we encounter in proof and term size versus complexity of the proof checker . reducing the proof size forces the checker ( the tcb ) to become more complex .
when considering the formal transformation , we start from a modified version of program [ core - twelf ] that makes all quantifiers explicit . to illustrate
, we apply the transformation to all of the declarations in the twelf encoding except for the constants and inference rules for pairing . applying the transformation to these declarations , we get the @xmath0prolog type declarations and clauses in programs [ trans - types ] and [ trans - rules ] .
type lam ltm - > ltm - > ( ltm - > ltm ) - > ltm .
type app ltm - > ltm - > ltm - > ltm - > ltm .
type eq ltm - > ltm - > ltm - > ltm .
type imp ltm - > ltm - > ltm .
infixr imp 7 .
type forall ltm - > ( ltm - > ltm ) - > ltm .
type false ltm .
type refl ltm - > ltm - > ltm .
type beta ltm - > ltm - > ( ltm - > ltm ) - > ltm - > ltm .
type congr ltm - > ltm - > ltm - > ( ltm - > ltm ) - > ltm - > ltm - > ltm .
type imp_i ltm - > ltm - > ( ltm - > ltm ) - > ltm .
type imp_e ltm - > ltm - > ltm - > ltm - > ltm .
type forall_i ltm - > ( ltm - > ltm ) - > ( ltm - > ltm ) - > ltm .
type forall_e ltm - > ( ltm - > ltm ) - > ltm - > ltm - > ltm .
.... hasltype ( lam t1 t2 f ) ( tm ( t1 arrow t2 ) ) : - hasltype t1 tp , hasltype t2 tp , pi x\ ( hasltype x ( tm t1 ) = > hasltype ( f x ) ( tm t2 ) ) .
hasltype ( app t1 t2 f x ) ( tm t2 ) : - hasltype t1 tp , hasltype t2 tp , hasltype f ( tm ( t1 arrow t2 ) ) , hasltype x ( tm t1 ) . hasltype ( eq t x y ) ( tm form ) : - hasltype t tp , hasltype x ( tm t ) , hasltype y ( tm t ) .
hasltype ( a imp b ) ( tm form ) : - hasltype a ( tm form ) , hasltype b ( tm form ) . hasltype ( forall t a ) ( tm form ) : - hasltype t tp , pi x\ ( hasltype x ( tm t ) = > hasltype ( a x ) ( tm form ) ) .
hasltype false ( tm form ) .
hasltype ( refl t x ) ( pf ( eq t x x ) ) : - hasltype t tp , hasltype x ( tm t ) . hasltype ( beta t1 t2 f x ) ( pf ( eq t2 ( app t1 t2 ( lam t1 t2 f ) x ) ( f x ) ) ) : - hasltype t1 tp , hasltype t2 tp , pi x\ ( hasltype x ( tm t1 ) = > hasltype ( f x ) ( tm t2 ) ) .
hasltype ( congr t x z h p1
p2 ) ( pf ( h x ) ) : - hasltype t tp , hasltype x ( tm t ) , hasltype z ( tm t ) , pi x\ ( hasltype x ( tm t ) = > hasltype ( h x ) ( tm form ) ) , hasltype p1 ( pf ( eq t x z ) ) , hasltype p2 ( pf ( h z ) ) .
hasltype ( imp_i a b q ) ( pf ( a imp b ) ) : - hasltype a ( tm form ) , hasltype b ( tm form ) . pi p\ ( hasltype p ( pf a ) = > hasltype ( q p ) ( pf b ) )
. hasltype ( imp_e a b q1 q2 ) ( pf b ) : - hasltype a ( tm form ) , hasltype b ( tm form ) , hasltype q1 ( pf ( a imp b ) ) , hasltype q2 ( pf a ) .
hasltype ( forall_i t a q ) ( pf ( forall t a ) ) : - hasltype t tp , pi y\ ( hasltype y ( tm t ) = > hasltype ( a y ) ( tm form ) ) , pi y\ ( hasltype y ( tm t ) = > hasltype ( q y ) ( pf ( a y ) ) ) .
hasltype ( forall_e t a q y ) ( pf ( a y ) ) : - hasltype t tp , pi y\ ( hasltype y ( tm t ) = > hasltype ( a y ) ( tm form ) ) , hasltype q ( pf ( forall t a ) ) , hasltype y ( tm t ) . .... before discussing the details , it is already possible to see some of the similarities between the twelf and @xmath0prolog encodings , and between the @xmath0prolog encoding resulting from the transformation and the one in programs [ core - types ] and [ core - rules ] .
for example , in twelf the full version of the ` congr ` rule is the ` congr ` proof constructor takes 6 arguments ( ` t ` , ` x ` , ` z ` , ` h ` , and two subproofs ) . in the @xmath0prolog version of ` congr ` in programs [ trans - types ] and [ trans - rules ] , ` congr ` also takes 6 arguments ( 4 terms and 2 subproofs ) though their types are different from the lf version .
also , in our original @xmath0prolog encoding ( program [ core - rules ] ) , the ` congr ` clause has 4 subgoals , while in the new one ( program [ trans - rules ] ) there are 6 ; it is easy to see the correspondence between 4 of them in the two encodings . note that in the version in program [ core - rules ] , two of them are typechecking subgoals and two are proof checking subgoals . in twelf , typechecking and proof checking
are unified , so all subgoals in the twelf version are twelf typechecking goals ; in our example some of them check terms whose types have the form ` ( tm a ) ` , while others check terms whose types have the form ` ( pf a ) ` . in lf , there are several kinds of assertions .
the two that are important for the formal transformation are : `` @xmath15 is a type '' and `` term @xmath22 has type @xmath15 '' .
two @xmath0prolog types ` ltp ` and ` ltm ` introduced in program [ trans - types ] are used to encode lf types and terms .
the @xmath0prolog predicates ` ltype ` and ` hasltype ` are introduced to express the two assertions , respectively .
the first assertion is important for transforming the three declarations in program [ core - twelf ] that end in `` ` type . ` '' they declare constants that are used to create lf types , which correspond to @xmath0prolog formulas ( terms of type ` o ` ) .
the second assertion is used for the rest . in order for an assertion of the second kind to hold
, it must also be the case that @xmath15 is a type .
for this reason , the @xmath0prolog predicate ` well_typed ` is included ( program [ trans - types ] ) and has one clause ( program [ trans - rules ] ) .
the declarations and clause discussed so far are necessary no matter what twelf encoding we begin with .
the remaining declarations and clauses in programs [ trans - types ] and [ trans - rules ] are specific to our particular object logic . for each twelf declaration in program [ core - twelf ] that we consider , there is one type declaration in program [ trans - types ] and one clause in program [ trans - rules ] .
the first change we make to the @xmath0prolog code in programs [ trans - types ] and [ trans - rules ] to get closer to an optimized version involves the ` well_typed ` clause .
consider the first subgoal of this clause , an ` ltype ` subgoal .
note that for our particular encoding , there are three clauses for the ` ltype ` predicate .
they correspond to the three kinds of objects in the encoding of the object logic : types , terms , and proofs . in solving an ` ltype ` subgoal , at most one clause will ever apply at any point depending on which of three forms the argument has .
this observation permits us to replace ` well_typed ` with the following three clauses which cover every case . in the first clause
, we can eliminate the ` ltype ` subgoal because it is always provable . in the second and third clauses
, we can replace the ` ltype ` subgoal with the corresponding subgoal from the body of the only ` ltype ` clause that applies , to obtain the clauses below .
although ` hasltype ` is sufficient for representing any lf assertion of the form `` term @xmath22 has type @xmath15 , '' in our encoding it is useful to distinguish three ways in which it is used .
this fact leads to our second modification of programs [ trans - types ] and [ trans - rules ] .
the second argument to ` hasltype ` always has one of the following forms : ` tp ` , ` ( tm t ) ` , or ` ( pf a ) ` . using this fact
, we replace ` hasltype ` with three predicates : ` istype ` , ` hastype ` , and ` proves ` .
since the second argument to ` istype ` always is ` tp ` , we can eliminate this argument altogether so that ` istype ` has type ` ltm - > o ` .
program [ trans - mod1 ] illustrates the modifications discussed so far on a subset of the ` hasltype ` clauses in program [ trans - rules ] , which include only those for ` arrow ` , ` forall ` , and ` forall_i ` . looking back at program [ core - twelf ] ,
note the types of the four constants that are used to construct terms of type ` tp ` .
there are no dependent types here ; they are all simple types , which could be transformed directly to @xmath0prolog types .
this fact leads to our third modification .
instead of transforming all twelf terms and types to @xmath0prolog terms as is done by the transformation , we transform types with no dependencies directly to @xmath0prolog types , thus allowing the @xmath0prolog typechecker to do more typechecking work automatically .
this direct transformation gives us the @xmath0prolog declarations this change forces several other changes .
the type of ` tm ` must be changed to ` tp - > ltp ` .
the ` well_typed ` clause for ` tp ` is no longer necessary .
the ` istype ` predicate and all of the clauses for it can be removed ; all ` istype ` subgoals in other clauses can be eliminated .
the ` well_typed ` clause for ` tm ` can also be eliminated since checking for well - typedness amounts to simply using the ` hastype ` predicate . in the types of all of the constants , wherever there appears a term ` t ` of type ` ltm ` such that ` t ` represents an object - logic type , the type of ` t ` must be changed to ` tp ` .
program [ trans - mod2 ] illustrates these changes on the subset of declarations and clauses from program [ trans - mod1 ] .
note that the types of ` forall ` and ` forall_i ` are changed to reflect the fact that the first argument ` t ` has type ` tp ` .
our fourth modification to the @xmath0prolog code allows the @xmath0prolog type system to make further useful distinctions for our particular object logic .
we introduced the ` hastype ` and ` proves ` predicate for the cases when the second argument to our old ` hasltype ` had the forms ` ( tm t ) ` and form ` ( pf a ) ` , respectively .
we can further simplify these clauses by eliminating the ` tm ` and ` pf ` constants . simply eliminating them
means we must change the types of the second argument to these predicates appropriately , but we can go a step further than that .
notice that after removing ` tm ` and ` pf ` , terms appear as the first argument to ` hastype ` and types as the second , and that proofs appear as the first argument to the ` proves ` predicate and formulas , which are a subset of the terms , appear as the second . to make these distinctions in the program ,
we reintroduce the constants ` tm ` and ` pf ` , but this time as @xmath0prolog types which replace ` ltm ` . after making all the changes discussed so far to the types and clauses in programs [ trans - types ] and [ trans - rules ] , we obtain the somewhat simpler versions in programs [ trans - types - a ] and [ trans - rules - a ] . type lam tp - > tp - > ( tm - > tm ) - > tm . type app tp - > tp - > tm - > tm - > tm .
type eq tp - > tm - > tm - > tm .
type imp tm - > tm - > tm .
infixr imp 7 .
type forall tp - > ( tm - > tm ) -
> tm . type false tm .
type refl tp - > tm - > pf .
type beta tp - > tp - > ( tm - > tm ) -
> tm - > pf . type congr tp - > tm - > tm - > ( tm - > tm ) - > pf - > pf - > pf .
type imp_i tm - > tm - > ( pf - > pf ) - > pf . type imp_e tm - > tm - > pf - > pf - > pf .
type forall_i tp - > ( tm - > tm ) - > ( tm - > pf ) - > pf .
type forall_e tp - > ( tm - > tm ) - > pf - > tm - > pf . .... hastype ( lam t1 t2 f ) ( t1 arrow t2 ) : - pi x\ ( hastype x t1 = > hastype ( f x ) t2 ) .
hastype ( app t1 t2 f x ) t2 : - hastype f ( t1 arrow t2 ) , hastype x t1 . hastype ( eq t x y ) form : - hastype x t , hastype y t. hastype ( a imp b ) form : - hastype a form , hastype b form .
hastype ( forall t a ) form : - pi x\ ( hastype x t = > hastype ( a x ) form ) .
hastype false form .
proves ( refl t x ) ( eq t x x ) : - hastype x t. proves ( beta t1 t2 f x ) ( eq t2 ( app t1 t2 ( lam t1 t2 f ) x ) ( f x ) ) : - pi x\ ( hastype x t1 = > hastype ( f x ) t2 ) . proves
( congr t x z h p1
p2 ) ( h x ) : - hastype x t ,
hastype z t , pi x\ ( hastype x t = > hastype ( h x ) form ) , proves p1 ( eq t x z ) , proves p2 ( h z ) . proves ( imp_i a b q ) ( a imp b ) : - hastype a form , hastype b form .
pi p\ ( proves p a = > proves ( q p ) b ) . proves ( imp_e a b q1 q2 )
b : - hastype a form , hastype b form , proves q1 ( a imp b ) , proves q2 a. proves ( forall_i t a q ) ( forall t a ) : - pi y\ ( hastype y t = > hastype ( a y ) form ) .
pi y\ ( hastype y t = > proves ( q y ) ( a y ) ) . proves ( forall_e t a q y ) ( a y ) : - pi y\ ( hastype y t = > hastype ( a y ) form ) , proves q ( forall t a ) , hastype y t. .... note that ` tm ` and ` pf ` no longer appear in clauses ( program [ trans - rules - a ] ) , and instead appear in types ( program [ trans - types - a ] ) . also note the new type and clause for ` well_typed ` as compared to what they were in program [ trans - mod2 ] . the types and clauses in programs [ trans - types - a ] and [ trans - rules - a ] are now quite close to those of programs [ core - types ] and [ core - rules ] in section [ core ] .
the remaining changes are optimizations that can be best illustrated if we view the @xmath0prolog code as a proof checker .
in particular , for any subgoal of the form ` ( proves p a ) ` , we assume the proof and the formula are given at the outset ( no logical variables ) and that the subgoal ` ( hastype a form ) ` will be asked first ( e.g. , via the ` well_typed ` predicate ) .
with this in mind , by looking at some of the clauses for the ` proves ` predicate , we find two kinds of redundancy .
consider , for example , the clause for ` refl ` .
the arguments ` t ` and ` x ` appear in both the proof and the formula . assuming that a formula and proof are always paired together , any arguments that appear in the formula do not have to be repeated in the proof .
thus we can remove both arguments to ` refl ` .
also , since we assume that the formula has already been typechecked , the ` hastype ` subgoal is redundant and can be eliminated .
thus we achieve the simple form for the ` refl ` rule as it appears in program [ core - rules ] .
next consider the clause for ` imp_e ` . since ` b ` is the formula whose proof is to be checked
, we do nt need an extra copy among the arguments to ` imp_e ` .
we also do nt need to typecheck ` b ` since this has been done via the initial call to ` well_typed ` .
if we are to guarantee correct typing of the formula in _ any _ ` proves ` subgoal generated during proof checking , then we need to keep ` hastype ` subgoals for any formula that does not appear as a subformula of the formula in the head of the clause . in the ` imp_e ` clause ,
the goal ` ( hastype a form ) ` is asked before ` ( proves q2 a ) ` and this ` hastype ` subgoal can not be removed .
these changes lead to the ` imp_e ` clause in program [ core - rules ] .
analogously , we can examine the ` hastype ` clauses and remove redundant arguments from terms .
for example , in the case of ` app ` , the type ` t2 ` can be removed because it appears as the second argument to ` hastype ` .
we must keep ` t1 ` if we want to preserve the property that proof checking will not introduce logic variables .
note that when comparing program [ trans - rules - a ] to program [ core - rules ] , in the ` proves ` clause for ` congr ` , no arguments are removed from the proof term in either case , even though ` h ` and ` x ` appear in the second argument to ` proves ` .
the reason is that backchaining on this clause requires higher - order matching , for which there can be more than one solution .
one further criteria that we place on our proof checker is that it can not backtrack .
thus we must include ` h ` and ` x ` explicitly in the proof term to prevent the possibility that when backchaining on this clause , a backtrack point is created by unification .
we can , however , eliminate the typechecking subgoal for ` h ` because its well - typedness follows from the fact that ` ( h x ) ` has type ` form ` and ` x ` has type ` t ` .
eliminating this subgoal from the clause in program [ trans - rules - a ] gives us the clause in program [ core - rules ] .
after making analogous changes to all of the clauses in program [ trans - rules - a ] , the only remaining difference in program [ core - rules ] is the use of ` assump ` to identify assumptions added during proof checking , which as stated earlier , is not necessary , but is useful for various programming tasks in our proof - carrying code system .
note that in making changes to the @xmath0prolog code , we have been careful not to complicate proof checking by requiring any more power from @xmath0prolog than was needed to execute the code obtained directly from the transformation .
the same is not true for the twelf code . as stated earlier , the version that used abbreviations ( program [ core - twelf ] ) needs more type reconstruction power than the version with all arguments explicitly included . in summary
, using the formal correspondence has provided a principled way to arrive at the versions of the encodings of the object logic in twelf ( program [ core - twelf ] ) and @xmath0prolog(program [ core - rules ] ) that we have compared .
the main differences are ( 1 ) the twelf encoding is more concise because dependent types eliminate the need for explicit typechecking subgoals , and ( 2 ) in @xmath0prolog , unlike twelf , proof checking of the optimized version of the encoding is no more complex than proof checking the original .
andrew w. appel and amy p. felty . a semantic model of types and machine instructions for proof - carrying code .
in _ the 27th annual acm sigplan - sigact symposium on principles of programming languages _ , pages 243253 , 2000 .
gopalan nadathur and dale miller .
an overview of @xmath0prolog . in robert
a. kowalski and kenneth a. bowen , editors , _ fifth international conference and symposium on logic programming _ , pages 810827 . mit press , 1988 .
gopalan nadathur and dustin .
j. mitchell .
system description : teyjus a compiler and abstract machine based implementation of @xmath0prolog . in _ the 16th international conference on automated deduction _ ,
pages 287291 .
springer - verlag , july 1999 .
gopalan nadathur and debra sue wilson . a representation of lambda terms suitable for operations on their intensions . in _ proceedings of the 1990 acm conference on lisp and functional programming _ , pages 341348 , 1990 .
frank pfenning and carsten schrmann .
system description : twelf a meta - logical framework for deductive systems . in _ the 16th international conference on automated deduction _ ,
pages 202206 .
springer - verlag , july 1999 .
|
@xmath0prolog is known to be well - suited for expressing and implementing logics and inference systems .
we show that lemmas and definitions in such logics can be implemented with a great economy of expression .
we encode a higher - order logic using an encoding that maps both terms and types of the object logic ( higher - order logic ) to terms of the metalanguage ( @xmath0prolog ) .
we discuss both the terzo and teyjus implementations of @xmath0prolog .
we also encode the same logic in twelf and compare the features of these two metalanguages for our purposes .
|
the mechanism of exclusive production of mesons at high energies became recently a very active field of research ( see @xcite and refs . therein ) . these reactions @xmath7 ,
where @xmath8 , @xmath2 , provide a valuable tool to investigate in detail the properties of resonance states at high energies .
the recent works concentrated on the production of @xmath9 mesons ( see e.g. refs .
@xcite ) where the qcd mechanism is similar to the exclusive production of the higgs boson . furthermore
, the @xmath10 states are expected to annihilate via two - gluon processes into light mesons and may , therefore , allow the study of glueball production dynamics .
the two - pion background to exclusive production of @xmath11 @xcite and @xmath2 @xcite mesons was already discussed . in ref .
@xcite a new perturbative mechanism of the @xmath12 production was discussed . due to reasons explained in ref .
@xcite this mechanism gives relatively small contribution in the @xmath2 invariant mass region .
the cdf collaboration has measured the cross section of cep @xmath9 mesons in proton - antiproton collisions at the tevatron @xcite . in this experiment @xmath9
mesons are identified via decay to the @xmath13 with @xmath14 channel .
the experimental invariant mass resolution was not sufficient to distinguish between scalar , axial and tensor @xmath9 . while the branching fractions to this channel for axial and tensor mesons are large @xcite ( @xmath15 and @xmath16 , respectively ) the branching fraction for the scalar meson is very small @xmath17 @xcite . on the other hand ,
the cross section for exclusive @xmath2 production obtained within the @xmath3-factorization is much bigger than that for @xmath18 and @xmath19 . as a consequence ,
all @xmath20 mesons give similar contributions @xcite to the @xmath13 decay channel .
clearly , the measurement via decay to the @xmath13 channel can not provide cross section for different @xmath9 .
the @xmath2 meson decays into several two - body channels ( e.g. @xmath21 , @xmath22 , @xmath23 ) or four - body hadronic modes ( e.g. @xmath24 , @xmath25 ) .
we have analyzed a possibility to measure @xmath2 via its decay into @xmath26 channel @xcite .
the branching fraction @xmath27 is large , the axial @xmath18 does not decay to the @xmath21 channel and @xmath28 is smaller @xcite .
in addition a much smaller cross section for @xmath19 production means that in practise only @xmath2 will contribute to the signal .
the advantage of this channel is that the @xmath26 continuum has been studied recently @xcite and is relatively well known .
the qcd mechanism of exclusive diffractive production of @xmath2 meson including the absorptive correction.,scaledwidth=25.0% ] the qcd mechanism for the diffractive production of heavy central system has been proposed by khoze , martin and ryskin ( kmr ) and developed in collaboration with kaidalov and stirling for higgs production ( see e.g. refs .
@xcite ) . in the framework of this approach the amplitude of the exclusive @xmath29 process
is described by the diagram shown in fig .
[ fig : fig1 ] , where the hard subprocess @xmath30 is initiated by the fusion of two off - shell gluons and the soft part is represented in terms of the off - diagonal unintegrated gluon distribution functions ( ugdfs ) .
the formalism used to calculate the exclusive @xmath2 meson production is explained in detail elsewhere @xcite . the double - diffractive mechanism of exclusive production of @xmath31 pairs including the pion and regge exchanges , the absorptive corrections due to proton - proton interactions as well as pion - pion rescattering.,title="fig:",scaledwidth=20.0% ] the double - diffractive mechanism of exclusive production of @xmath31 pairs including the pion and regge exchanges , the absorptive corrections due to proton - proton interactions as well as pion - pion rescattering.,title="fig:",scaledwidth=20.0% ] the dominant mechanism of the exclusive production of @xmath31 pairs at high energies is sketched in fig . [ fig : central_double_diffraction_diagrams_fsi ] .
the expected non - resonant background can be modeled using a `` non - perturbative '' framework where the pion pair is produced by pomeron - pomeron fusion with an intermediate off - shell pion / reggeon exchanged between the final - state particle pairs ( see @xcite for details ) . in calculations of the amplitude
we follow the general rules of pumplin and henyey @xcite used recently in ref .
@xcite where a first estimate of the differential cross sections for the @xmath26 pairs production at the lhc energies has been presented .
the regge parametrization of the @xmath32 and @xmath33 scattering amplitude includes both pomeron as well as @xmath34 and @xmath35 reggeon exchanges with the parameters taken from the donnachie - landshoff analysis @xcite of the total cross sections .
the regge - type interaction applies at higher energies and at low energies should be switched off ( see @xcite ) . in ref .
@xcite we propose to use a generalized propagator .
the form factors correct for the off - shellness of the intermediate pions are parametrized as @xmath36 , where the parameter @xmath37 is obtained from fit to the experimental data @xcite ( see @xcite ) .
we first show ( fig . [
fig : diff_comp ] ) the differential cross sections of @xmath2 cep at @xmath38 = 14 tev without ( dashed line ) and with ( solid line ) absorptive corrections .
these calculations were done with gjr nlo @xcite collinear gluon distribution , to generate the kmr ugdfs , which allows to use low values of the internal gluon transverse momenta @xmath39 gev@xmath40 .
the bigger the value of the cut - off parameter , the smaller the cross section ( see ref .
@xcite ) . in the calculations
we take the value of the hard scale to be @xmath41 .
the smaller @xmath42 , the bigger the cross section @xcite . in all cases
the absorption effects lead to a damping of the cross section . in most cases
the shape is almost unchanged .
exception is the distribution in proton transverse momentum where the absorption effects lead to a damping of the cross section at small proton @xmath43 and an enhancement of the cross section at large proton @xmath43 . in relative azimuthal angle distribution
we observe a dip at @xmath44 .
transverse momentum distribution of @xmath2 shows a small minimum at @xmath45 2.5 gev .
the main reason of its appearance is the functional dependence of matrix elements on its arguments @xcite . + in fig .
[ fig : dsig_dmpipi ] we compare differential distributions of pions from the @xmath2 decay ( see the peak at @xmath46 gev ) with those for the continuum pions .
while left panels show the cross section integrated over the full phase space , the right panels show results including the relevant pion pseudorapidities restrictions @xmath47 ( rhic and tevatron ) and @xmath48 ( lhc ) .
the @xmath2 contribution is calculated with grv94 nlo @xcite and gjr08 nlo @xcite collinear gluon distributions . in fig .
[ fig : pt_lhc ] we show distributions in pion transverse momenta ( left panels ) .
the pions from the @xmath2 decay are placed at slightly larger transverse momenta .
this can be therefore used to get rid of the bulk of the continuum by imposing an extra cut on the pion transverse momenta . in the right panels we show two - pion invariant mass distributions with additional cuts on both pion transverse momenta @xmath49 gev .
now the signal - to - background ratio is somewhat improved especially at the tevatron and lhc energies .
the main experimental task is to measure the distributions in the @xmath2 rapidity and transverse momentum . in fig .
[ fig : ratio ] we show the two - dimensional ratio of the cross sections for the @xmath2 meson in its rapidity and transverse momentum : @xmath50 the numerator includes limitations on @xmath51 and @xmath52 .
these distributions provide a fairly precise evaluation of the expected acceptances when experimental cuts are imposed .
the experimental data could be corrected by our two - dimensional acceptance function to recover the distributions of interest .
it was realized recently that the measurement of exclusive production of @xmath20 via decay in the @xmath13 channel can not give production cross sections for different species of @xmath20 . in this decay channel
the contributions of @xmath20 mesons with different spins are similar and experimental resolution is not sufficient to distinguish them .
we have analyzed a possibility to measure the exclusive production of @xmath2 meson in the proton-(anti)proton collisions at the lhc , tevatron and rhic via @xmath53 decay channel .
since the cross section for exclusive @xmath2 production is much larger than that for @xmath18 and @xmath19 and the branching fraction to the @xmath21 channel for @xmath2 is larger than that for @xmath19 ( @xmath18 does not decay into two pions ) the two - pion channel should provide an useful information about the @xmath2 cep .
we have performed detailed studies of several differential distributions and demonstrated how to impose extra cuts in order to improve the signal - to - background ratio .
the two - pion background was calculated in a simple model with parameters adjusted to low energy data ( see @xcite ) .
we have shown that relevant measurements at tevatron and lhc are possible . at rhic
the signal - to - background ratio is much worse but measurements should be possible as well . imposing cuts distorts the original distributions for @xmath2 in rapidity and transverse momentum .
we have demonstrated how to recover the original distributions and presented the correction functions for some typical experimental situations .
khoze , a.d .
martin and m.g .
ryskin , phys . lett . *
b401 * ( 1997 ) 330 ; v.a .
khoze , a.d .
martin and m.g .
ryskin , eur .
j. * c23 * ( 2002 ) 311 ; a.b .
kaidalov , v.a .
khoze , a.d .
martin and m.g .
ryskin , eur .
j. * c33 * ( 2004 ) 261 ; a.b .
kaidalov , v.a .
khoze , a.d .
martin and m.g .
ryskin , eur .
j. * c31 * ( 2003 ) 387 ; v.a .
khoze , a.d .
martin , m.g .
ryskin and w.j .
stirling , eur .
j. * c35 * ( 2004 ) 211 .
|
we report on the results of a theoretical study of the central exclusive production of scalar @xmath0 meson via @xmath1 decay in high - energy hadron collisions at the rhic , tevatron and lhc .
the corresponding amplitude for exclusive double - diffractive @xmath2 meson production was obtained within the @xmath3-factorization approach including virtualities of active gluons and the cross section is calculated with unintegrated gluon distribution functions ( ugdfs ) known from the literature .
the four - body @xmath4 reaction constitutes an irreducible background to the exclusive @xmath2 meson production .
we include the absorption effects due to proton - proton interaction and pion - pion rescattering .
several differential distributions for @xmath5 process , including the absorptive corrections , were calculated .
the influence of kinematical cuts on the signal - to - background ratio is investigated .
@xmath6 decay , diffractive processes , two - pion continuum
|
maximally random jammed ( mrj ) packings of hard particles are the most disordered structures , according to some well - defined order metrics , that are rigorously incompressible and nonshearable @xcite .
these systems are prototypical `` glassy '' structures in the sense that they are structurally rigid yet lack bragg peaks in their scattering spectra @xcite . in this sense
, the idea of the mrj state has replaced the mathematically ill - defined notion of random close packing @xcite .
nearly half - a - century ago these systems were thought to describe the disordered structure of liquids @xcite , but it is now known that three - dimensional mrj monodisperse sphere packings possess unusual quasi - long - range ( qlr ) pair correlations decaying as @xmath0 @xcite .
this property is markedly different from typical liquids , in which pair correlations decay exponentially fast @xcite .
similar qlr behavior has also been observed in noninteracting spin - polarized fermionic ground states @xcite , the ground state of liquid helium @xcite , and the harrison - zeldovich power spectrum of the density fluctuations of the early universe @xcite .
however , for each of these examples and for mrj hard sphere packings , the structural origins of these correlations have been heretofore unknown , even for monodisperse systems .
furthermore , it is an open problem to generalize these qlr correlations for mrj states to polydisperse packings , in which the jamming properties are intimately related to the size - distribution of the particles @xcite .
motivated by the observation that mrj packings are structurally rigid with a well - defined contact network , torquato and stillinger conjectured @xcite that all strictly jammed ( i.e. , mechanically rigid ) , saturated @xcite packings of monodisperse spheres in @xmath2-dimensional euclidean space @xmath4 are hyperuniform , meaning that infinite - wavelength local density fluctuations vanish @xcite , a proposition for which no counterexample has been found to date @xcite .
this conjecture suggests that saturation and strict jamming are sufficient to induce hyperuniformity , albeit not necessary @xcite .
hyperuniform systems play an integral role in understanding the relationship between fluctuations in local material properties and microstructural order @xcite .
these systems have applications to the large - scale structure of the universe @xcite , the structure and collective motion of grains in vibrated granular media @xcite , the structure of living cells @xcite , transport through composites and porous media @xcite , the study of noise and granularity of photographic images @xcite , identifying properties of organic coatings @xcite , and the fracture of composite materials @xcite . for microstructures consisting of `` point '' particles , one considers fluctuations in the local number density within some observation window .
hyperuniform point patterns possess local density fluctuations that asymptotically grow more slowly than the volume of the window .
recently , the concept of hyperuniformity has been extended to include systems composed of finite - volume inclusions of arbitrary geometries @xcite ; in these cases , the quantity of interest is the fluctuation in the so - called local volume fraction , defined as the fraction of the volume within an observation window covered by a given phase .
hyperuniform heterogeneous media possess local - volume - fraction fluctuations that asymptotically decay faster than the volume of the observation window , implying that the local volume fraction approaches a global value beyond relatively few characteristic length scales .
previous work on three - dimensional ( 3d ) mrj monodispere sphere packings @xcite has supported the torquato - stillinger conjecture by showing that the structure factor @xmath5 , proportional to the scattering intensity , approaches zero linearly as the wavenumber @xmath6 , inducing a qlr power - law tail @xmath0 in the pair correlation function @xmath7 .
this behavior implies that local - number - density fluctuations grow logarithmically faster than the surface area of an observation window but still slower than the window volume . in this sense ,
the `` degree '' of hyperuniformity in mrj packings is minimal among all strictly jammed saturated packings .
however , recent numerical @xcite and experimental @xcite work on polydisperse mrj packings has suggested that hyperuniform quasi - long - range correlations are not a universal signature of the mrj state . unlike monodisperse systems
, the distribution of particle sizes in a polydisperse packing introduces locally inhomogeneous regions as the particles distribute themselves through space as shown in figure [ figone ] .
@xmath8{fig1a } & \includegraphics[width=0.45\textwidth]{fig1b}\\ \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ] one can see in the binary packing that local clusters of small particles are distributed near and around larger inclusions , and the result is that the point pattern generated by the disk centers possesses local inhomogeneities that are expected to ( and indeed do ) induce volume - order scaling within the number variance .
the situation is apparently even more complex for the polydisperse system in figure [ figone ] since the size distribution results in a highly inhomogeneous local structure with small particles trapped between larger ones with a high probability .
these observations have raised a number of quantitatively and conceptually difficult questions .
first , what is the appropriate extension of the torquato - stillinger conjecture for monodisperse mrj packings to systems with a size distribution ? clearly , one must explicitly account for the shape information of the particles .
second , in the event that one can generalize the torquato - stillinger conjecture , there is to date no satisfactory structural explanation for the _ linear _ small - wavenumber scaling of the structure factor observed for 3d mrj monodisperse hard sphere packings , which indicates the presence of qlr pair correlations .
this extraordinarily difficult problem is tantamount to providing an analytical prediction of the mrj state .
unfortunately , no such rigorous theory currently exists , even for the considerably simpler problem of predicting the scalar mrj density @xcite .
the presence of qlr correlations makes this problem inherently nonlocal , and , therefore , methods that attempt to predict the mrj state based only on packing fraction and local criteria , such as nearest - neighbor and voronoi statistics , are invariably incomplete @xcite .
we have presented arguments in a recent letter to suggest strongly that hyperuniformity and quasi - long - range pair correlations are signatures of saturated mrj packings of hard particles , including binary disks , ellipses , and superdisks @xcite . in this paper , we provide detailed evidence to show that polydisperse mrj packings of hard spheres , though inhomogeneous with respect to the number variance , possess local - volume - fraction fluctuations decaying faster than the volume of an observation window .
this observation is consistent with the generalized torquato - stillinger conjecture that all strictly jammed saturated packings of spheres are hyperuniform with respect to fluctuations in the _ local volume fraction_. our major results include : 1 .
infinite wavelength local number density fluctuations do not vanish for mrj packings of polydisperse hard spheres .
local - volume - fraction fluctuations provide the appropriate structural description of these packings because they account correctly for the size distribution of the particles .
importantly , our work studying local - volume - fraction fluctuations contains previously - published results for 3d monodisperse mrj hard sphere packings as a special case .
2 . signature qlr pair correlations scaling asymptotically as @xmath1 in @xmath2 euclidean dimensions are observed for all systems that we study , including binary disks with varying size ratios and compositions and polydisperse disk packings .
our results suggest that these special correlations may be a universal feature of the mrj state .
3 . strict jamming places a strong constraint on the distribution of the available void space external to the particles such that hyperuniformity is observed when considering local - volume - fraction fluctuations , even when the point pattern of the sphere centers is locally inhomogeneous .
the competition between maximal randomness and strict jamming of the packings ensures that the void - space distribution is sufficiently broad to induce qlr correlations between particles , thereby providing a direct qualitative _ structural _ explanation for the linear small - wavenumber region of the generalized scattering intensity .
we consider point patterns to be realizations of stochastic point processes .
formally , a _
stochastic point process _ is a method of placing points in some space ( such as @xmath4 ) according to an underlying probability distribution .
this random setting is quite general , incorporating cases in which the locations of the points are deterministically known , such as in a bravais lattice @xcite .
a statistically homogeneous point process is completely determined by the number density @xmath9 and the countably infinite set ( in the thermodynamic limit ) of _ @xmath3-particle correlation functions_. the @xmath3-particle correlation function @xmath10 is proportional to the probability density of finding @xmath3 particles in volume elements around the positions @xmath11 , regardless of the positions of the remaining particles in the system . for an arbitrary point process ,
deviations of @xmath12 from unity provide a measure of the correlations among points in the system . note that specifying only a finite number @xmath13 of the @xmath3-particle correlation functions defines
a class of microstructures with degenerate @xmath13-particle statistics @xcite . of particular interest
is the pair correlation function @xmath14 , which defines the average number of particles surrounding a reference particle of the point process .
closely related to the pair correlation function is the _ total correlation function _ @xmath15 . since @xmath16 as @xmath17 ( @xmath18 ) for isotropic ,
translationally invariant systems without long - range order , it follows that @xmath19 in this limit , meaning that @xmath20 is generally integrable with a well - defined fourier transform .
it is common in statistical mechanics when passing to reciprocal space to consider the associated _ structure factor _ @xmath5 , which for a translationally invariant system is defined by @xmath21 where @xmath22 is the fourier transform of the total correlation function , @xmath9 is the number density , and @xmath23 is the magnitude of the reciprocal variable to @xmath24 .
we utilize the following definition of the fourier transform : @xmath25 where @xmath26 is the conventional euclidean inner product of two real - valued vectors . for radially - symmetric functions [ i.e. , @xmath27
, the fourier transform may be written @xmath28 closely related to the notion of a stochastic point process is that of a two - phase random heterogeneous medium ( or random set ) , which we define to be a domain of space @xmath29 of volume @xmath30 that is composed of two regions : the phase 1 region @xmath31 of volume fraction @xmath32 and the phase 2 region @xmath33 of volume fraction @xmath34 @xcite .
the statistical properties of each phase of the system are specified by the countably infinite set of _ @xmath3-point probability functions _
@xmath35 , which are defined by @xcite @xmath36 where @xmath37 is the indicator function for phase @xmath38 @xmath39 the function @xmath40 defines the probability of finding @xmath3 points at positions @xmath41 all within the same phase . upon subtracting the long - range behavior from @xmath42 ,
one obtains the autocovariance function @xmath43 , which is generally integrable .
it is important to recognize that the autocovariance function is independent of the choice of reference phase , meaning that it is a _
global _ descriptor of correlations within the system .
this property will play a particularly important role in this paper when we consider the relationship between local - volume - fraction fluctuations and the void space between inclusions in the microstructure .
the analog of the structure factor in this context is the so - called _ spectral density _ , which is the fourier transform @xmath44 of the autocovariance function @xcite .
the autocovariance function obeys the bounds @xcite @xmath45 where @xmath46 is the volume fraction of an arbitrary reference phase .
we remark that it is an open problem to identify additional necessary and sufficient conditions that the autocovariance function must satisfy in order to correspond to a binary stochastic process @xcite .
a _ sphere packing _ is a collection of non - overlapping spheres in @xmath2-dimensional euclidean space @xmath4 .
the _ packing density _ @xmath46 ( equivalent to the volume fraction of the particle phase ) is defined as the fraction of space covered by the spheres , which may be polydisperse . an important characteristic of a packing is its degree of randomness ( or the antithesis , order ) , which reflects nontrivial information of the packing structure .
the degree of randomness ( order ) can be quantified by a set of _ order metrics _
@xmath47 @xcite .
it is an open and challenging problem to identify good order metrics , but it has recently been proposed that hyperuniformity is itself a measure of order on large length scales @xcite
. one method of classifying sphere packings involves characterizing the extent to which particles are _
jammed_. torquato and stillinger @xcite have provided a precise definition of the term _ jamming _ and have proposed a hierarchical classification scheme for sphere packings by invoking the notions of _ local _ , _ collective _ , and _ strict _ jamming .
a packing is _ locally _ jammed if no particle in the system can be translated while fixing the positions of all other particles .
a _ collectively _ jammed packing is locally jammed such that no subset of spheres can simultaneously be continuously displaced without moving its members out of contact both with one another and with the remainder set .
a packing is _ strictly _ jammed if it is collectively jammed and if all globally uniform volume nonincreasing deformations of the system boundary are disallowed by the impenetrability constraints .
the reader is referred to ref .
@xcite for further details .
as previously mentioned , the _ maximally random jammed _
( mrj ) state is defined as the most disordered jammed packing in a given jamming category ( i.e. , locally , collectively , or strictly jammed ) @xcite .
the mrj state is well - defined for a given jamming category and choice of order metric , and it has recently supplanted the ill - defined random close packed ( rcp ) state @xcite . in this paper , we focus on maximally random strictly jammed polydisperse sphere packings in @xmath4 .
a _ hyperuniform point process _ has the property that the variance in the number of points in an observation window @xmath48 grows more slowly than the volume of that window . in the case of a spherical observation window
, this definition implies that the local number variance @xmath49 grows more slowly than @xmath50 in @xmath2 dimensions , where @xmath51 is the radius of the observation window .
torquato and stillinger @xcite have provided an exact expression for the local number variance of a statistically homogeneous point process in a spherical observation window @xmath52,\ ] ] where @xmath51 is the radius of the observation window , @xmath53 is the volume of the window , and @xmath54 is the so - called _ scaled intersection volume_. the latter quantity is geometrically defined as the volume of space occupied by the intersection of two spheres of radius @xmath51 separated by a distance @xmath55 and normalized by the volume of a sphere @xmath53 .
exact expressions for @xmath54 in arbitrary dimensions have been given by torquato and stillinger @xcite .
it is convenient to introduce a dimensionless density @xmath46 , which need not correspond to the volume fraction , according to @xmath56 where @xmath57 is a characteristic length scale of the system ( e.g. , the mean nearest - neighbor distance between points ) .
the number variance admits the following asymptotic scaling @xcite : @xmath58\right\},\ ] ] where @xmath59 denotes all terms of order less than @xmath60 . this result is valid for all periodic point patterns ( including lattices ) , quasicrystals that possess bragg peaks , and disordered systems in which the pair correlation function @xmath14 decays to unity exponentially fast @xcite .
explicit forms for the asymptotic coefficients @xmath61 and @xmath62 are given by @xcite @xmath63 where @xmath64\}$ ] .
any system with @xmath65 satisfies the requirements for hyperuniformity .
although the expansion will hold for all periodic and quasiperiodic point patterns with bragg peaks , this behavior is not generally true for disordered hyperuniform systems .
for example , it is known that if the total correlation function @xmath66 for large @xmath55 [ @xmath67 for small @xmath68 , then @xmath69 @xcite .
such behavior occurs in maximally random jammed monodisperse sphere packings in three dimensions @xcite and noninteracting spin - polarized fermion ground states @xcite .
other examples of `` anomalous '' local density fluctuations have been characterized by zachary and torquato @xcite . in order to define hyperuniformity for heterogeneous media
we introduce the _ local volume fraction _
@xmath70 of phase @xmath38 according to @xmath71 where @xmath53 is the volume of the observation window and @xmath72 is the corresponding indicator function . using this definition , the variance @xmath73 in the local volume fraction
is given by @xmath74 which is independent of the choice of reference phase .
the variance in the local volume fraction admits the asymptotic expansion @xcite
@xmath75\right\}\label{tauasymp}\\ a_{\tau } & = \int_{\mathbb{r}^d } \chi(\mathbf{r } ) d\mathbf{r } = \lim_{\lvert\mathbf{k}\rvert\rightarrow 0 } \hat{\chi}(\mathbf{k})\label{atau}\\ b_{\tau } & = -\frac{\kappa(d)}{d } \int_{\mathbb{r}^d } \lvert \mathbf{r}\rvert \chi(\mathbf{r } ) d\mathbf{r}\label{btau}.\end{aligned}\ ] ] the coefficients @xmath76 and @xmath77 in and control the asymptotic scaling of the fluctuations in the local volume fraction .
it then follows that @xmath78 decays faster than @xmath79 as @xmath80 for those systems such that @xmath81 which defines a hyperuniform two - phase random heterogeneous medium @xcite .
here we consider how the presence of polydispersity in a sphere packing affects the fluctuations in the local number density and local volume fraction .
for a packing of polydisperse spheres , the distribution of sphere radii is determined by a probability density @xmath82 , and the fundamental statistical descriptors of the medium therefore involve averages @xmath83 @xcite over the distribution of sphere sizes @xcite . for a packing of polydisperse spheres with @xmath13 distinct radii , the density function @xmath82 takes the form @xmath84 where @xmath85 is the mole fraction of species @xmath38 . as an example
, we will continually refer in this paper to the average particle diameter @xmath86 of a polydisperse hard - disk packing : @xmath87 where @xmath88 is the diameter of species @xmath38 .
the autocovariance function @xmath89 for a heterogeneous medium consisting of impenetrable polydisperse spheres is given by @xcite @xmath90 which implies that @xmath91 unlike for monodisperse sphere packings @xcite , the result shows that it is generally not possible to separate the shape information of the inclusions from the details of the point pattern generated by the sphere centers .
this observation suggests that for polydisperse microstructures , hyperuniformity of the underlying point pattern does not induce hyperuniformity with respect to local - volume - fraction fluctuations ; conversely , it is also possible to find heterogeneous media for which @xmath92 but @xmath93 .
we note that the first term contributing to @xmath89 in can be interpreted as the probability of finding two points , separated by a displacement @xmath24 , in a single particle of the packing .
the second term is therefore related to the probability of finding the points in two different particles .
it is clear that only this latter term , containing the pair correlation function , can contribute to the linear small - wavenumber region of the spectral density , and it is therefore responsible , albeit in a highly nontrivial way , for the onset of qlr correlations in mrj packings .
motivated by the generalized conjecture that all strictly jammed packings of @xmath2-dimensional spheres are hyperuniform with respect to local - volume - fraction fluctuations , we have generated several packings of binary disks ( 2d ) near the mrj state using a modified lubachevsky - stillinger ( ls ) packing algorithm @xcite , wherein particles with a fixed size ratio undergo event - driven molecular dynamics while simultaneously increasing in size according to some prescribed growth rate .
the initial growth rate used in our simulations is @xmath94 but near the jamming point , a much smaller growth rate @xmath95 is used to establish well - defined interparticle contacts .
the lubachevsky - stillinger algorithm has been shown to produce mrj packings with these parameters consistent with the positively - correlated translational and orientational order metrics @xcite .
our statistics for binary packings are averaged over 50 configurations of 10000 particles ; for polydisperse packings we have averaged over 10 configurations of 10000 particles .
our results have been compared to systems of up to @xmath96 particles to verify invariance of the statistics with respect to system size . for this study , we have chosen a particle size ratio @xmath97 and mole fractions @xmath98 and @xmath99 ; however , the focus of this study is to elucidate a universal property of polydisperse mrj packings , and our results are expected to apply for a range of size ratios , size distributions , and mole fractions ( see figure [ figtwo ] ) @xcite . for the remainder of this paper , we therefore focus on the binary case with the disclaimer that our results are expected to apply for general polydisperse mrj packings .
this algorithm has been used in previous studies of 3d monodisperse sphere packings near the mrj state @xcite .
the resulting packings in this study have a final volume fraction @xmath100 , which is below the the close - packed density @xmath101 .
figure [ figone ] ( section i ) provides a typical realization of a binary packing ; note that the particle distribution is saturated and nonperiodic .
we have calculated both the structure factor @xmath5 and the spectral density @xmath102 for the binary mrj disk packings using discrete fourier transforms of the local density @xmath103 and indicator function @xmath104 of the particle phase of the packing .
specifically , @xmath105 where @xmath106 is the fourier transform of the indicator function for a @xmath2-dimensional sphere of radius @xmath51 @xcite . note that the shape of the enclosure ( defined by a set of basis vectors @xmath107 ) restricts the wavevectors such that @xmath108 for all @xmath38 , where @xmath109 .
since the zero wavevector is removed from the spectrum in the expressions above , one must define @xmath110 with a similar expression for @xmath111 .
the limit here is taken at constant number density @xmath112 , and @xmath113 is the smallest computable wavevector as determined by the shape of the boundary . to obtain radially - symmetric forms of the structure factor and spectral density , we radially average over all wavevectors with equal magnitude .
our results for the structure factors and spectral densities of the binary and polydisperse mrj hard disk packings are shown in figure [ figtwo ] .
@xmath114{fig2a } & \includegraphics[width=0.35\textwidth]{fig2b } \\
\mbox{\bf ( a ) } & \mbox{\bf ( b)}\\ \includegraphics[width=0.35\textwidth]{fig2c } & \includegraphics[width=0.35\textwidth]{fig2d } \\ \mbox{\bf ( c ) } & \mbox{\bf ( d ) } \end{array}$ ] the structure factors for these systems lacks bragg peaks , reflecting the absence of long - range order .
of particular importance is the behavior of the structure factor near the origin .
a fit of the small-@xmath115 region ( @xmath116 ) with a third - order polynomial suggests @xmath117 , meaning that the point pattern generated by the disk centers does not possess vanishing infinite - wavelength local number density fluctuations , an observation verified by direct calculation of the number variance in figure [ mrjnum ] .
note that the local number variance asymptotically scales with the volume of the observation window .
@xmath8{fig3a } & \includegraphics[width=0.45\textwidth]{fig3b } \\
\mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ] this behavior differs from mrj packings of _ monodisperse _ spheres in three dimensions @xcite , where the structure factor decays linearly to zero for small wavenumbers .
bidisperse distributions of disks are inherently _ inhomogeneous _ with respect to the locations of the particle centers . for the systems studied here , the large concentration of small particles results in local clusters that are essentially close - packed . however , the introduction of large particles into the system generates effective grain boundaries between these local clusters , breaking the uniformity of the underlying point pattern .
we note that the structure factor for our binary packing can be decomposed as @xmath118 where @xmath119\left[\sum_{\ell=1}^{n_l } \exp(i\mathbf{k}\cdot\mathbf{x}_\ell)\right]}{n}\right\}\end{aligned}\ ] ] are the particle structure factors incorporating small - small , large - large , and small - large correlations , respectively .
note that @xmath120 is the number of small particles with positions @xmath121 , and @xmath122 is the number of large particles with positions @xmath123 .
these partial contributions to the structure factor are shown in figure [ psk ] . , @xmath124 , and @xmath125 to the structure factor of the binary mrj disk packings.,scaledwidth=50.0% ] none of the partial contributions to the structure factor possess a vanishing small - wavenumber region , implying that the local density fluctuations of the small and large particles each scale with the volume of an observation window as does the covariance .
these observations clearly demonstrate that information contained in the structure factor is not sufficient to characterize the packings because it neglects the details of the particle shapes . a direct calculation of the spectral density for the binary mrj system shows markedly different behavior from the structure factor as seen in figure [ figtwo ] .
one notices that infinite - wavelength local - volume - fraction fluctuations are suppressed by the system .
we have fit the small-@xmath115 region of the spectral density ( in units of @xmath126 ) using a third - order polynomial of the form @xmath127 and have found @xmath128 strongly suggesting that this system is hyperuniform with respect to local - volume - fraction fluctuations .
we have verified this claim by directly calculating the variance in the local volume fraction , shown in figure [ mrjnum ] .
the local - volume - fraction fluctuations decay faster than the volume of an observation window and logarithmically slower than the surface area , consistent with the presence of qlr correlations .
the spectral density is dominated by a peak with wavelength just below @xmath86 , approximately corresponding to the small - particle diameter .
this observation is consistent with the microstructure of the medium , which contains several regions of almost close - packed clusters of small particles . in order to further understand the behavior of the spectral density ,
it is important to recognize that both the local distribution of inclusions and the shape information of the particles contribute to @xmath44 .
however , the relative influence of each component on the spectral density varies throughout the spectrum . to characterize this effect
, we recall the representation of the spectral density , which admits the following upper bound : @xmath129 figure [ mag ] plots the average @xmath130 , which according to controls the upper bound on the spectral density . of the spatial contribution to the spectral density.,scaledwidth=50.0% ]
note that for small wavenumbers [ @xmath131 the bound places only weak constraints on the spectral density , implying that this region of the spectrum is controlled by information in the local structure of the heterogeneous medium .
we emphasize that the shape information of the particles must still be included at small wavenumbers to account for vanishing infinite - wavelength local - volume - fraction fluctuations , but , as we show below , such geometric information only provides the appropriate weights for the partial contributions to the structure factor to induce hyperuniformity .
in constrast , the large - wavenumber region of the spectrum closely follows the behavior of the upper bound , meaning that the length scale imposed by the decoration of the particle centers , here chosen to be disks , controls the spectral density .
we emphasize that this portion of the spectrum arises from the shape information of the inclusions themselves and is almost entirely independent of the distribution of particles in the system . specifically ,
the particle indicator function @xmath132 , which has compact support @xmath133 $ ] , possesses a fourier representation @xmath134 that is both long - ranged and has an intrinsic period , which for a binary packing leads to interference effects in the large - wavenumber region .
similar behavior arises in monodisperse systems , where the spectral density is exactly given @xmath135 , and one can rigorously separate information contained in the point pattern generated by the sphere centers from the shape information of the inclusions . figure [ stepchi ] shows the spectral density and shape contribution for a system of impenetrable disks with pair correlation function @xmath136 .
note that the bound only applies when @xmath5 is at a maximum ; for clarity we have omitted the associated scaling factor on the shape contribution . process .
note that the shape contribution controls the spectral density for large wavenumbers.,scaledwidth=50.0% ] as with the binary mrj packing , this `` step - function process '' possesses a spectral density that for small wavenumbers depends signficantly on the distribution of sphere centers ; the large - wavenumber region is almost exactly equal to the shape contribution since the structure factor approaches an asymptotic value of unity .
although for binary systems polydispersity precludes the direct separation of the shape information from the underlying point pattern , these examples show that qualitatively these components continue to affect only specific portions of the spectral density .
the dominance of the shape information for large wavenumbers has the surprising effect of almost completely suppressing fluctuations for @xmath137 , thereby suggesting that local volume fraction fluctuations essentially vanish on this length scale .
in actuality , the nonuniformity of the microstructure precludes a complete extinction of the variance in the local volume fraction ; nevertheless , the local - volume - fraction variance ( figure [ mrjvar ] ) undergoes a sharp change in slope near this length scale , highlighting a transition to the `` geometrically - controlled '' region of the spectral density .
physically , this behavior implies that observation windows with radii given by the appropriate wavelength capture the effective pore size surrounding the inclusions in such a way that the medium is essentially homogeneous on this _ local _ scale .
this observation suggests that the void space surrounding the particles plays a central role in determining the spatial statistics of a heterogeneous system .
it is important to note that the small-@xmath115 behavior of the spectral density for the binary mrj packing results in an `` anomalous '' asymptotic scaling of the variance in the local volume fraction .
specifically , our results suggest that the spectral density is nonanalytic at the origin with an expansion @xmath138 as @xmath139 , implying that the autocovariance function exhibits quasi - long - range behavior and scales with @xmath140 [ @xmath1 in @xmath2 dimensions ] .
equivalent behavior has been observed for the structure factor of mrj monodisperse sphere packings in three dimensions @xcite , and the number variance in that case has been shown to scale according to : @xmath141 a related scaling must also hold for the variance in the local volume fraction of our binary mrj packings ; specifically : @xmath142 one can directly see the effect of this behavior in figure [ mrjvar ] , where the variance in the local volume fraction for the mrj packings is compared to the square lattice packing and its binary variant , the unit cells for which are shown in figure [ zbinz ] .
@xmath8{fig8a } & \includegraphics[width=0.45\textwidth]{fig8b}\\ \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ] note that local volume fraction fluctuations in the mrj packings decay more slowly than in either of the periodic systems ; indeed , as the size distribution of the void space becomes smaller and more uniform , local - volume - fraction fluctuations are more rapidly suppressed on the global scale of the microstructure .
since the integer packing is not even strictly jammed , this comparison suggests that strict jamming is neither a necessary nor a strong determinant of hyperuniformity in heterogeneous media .
identifying the origin of the linear small - wavenumber region of the spectral density is an open problem that must be related to the structural features of the mrj state .
this problem is particularly difficult for our polydisperse packings since the underlying point pattern generated by the particle centroids possesses non - vanishing infinite - wavelength local number density fluctuations .
however , the spectral density @xmath102 can be expressed in terms of the partial structure factors as @xmath143 where @xmath144 and @xmath145 are the radii of the small and large particles , respectively .
this result follows directly from an expansion of .
as @xmath139 , we therefore find @xmath146 implying that the particle volumes provide the appropriate _ weights _ to properly balance the small- and large - particle variances with the covariance between the particles .
additionally , since @xmath134 possesses no linear term in its small - wavenumber taylor expansion , it follows that the appearance a linear small - wavenumber region in the spectral density , and therefore qlr correlations , _ must _ involve an appropriate superposition of linear contributions in the partial structure factors .
this observation suggests that quasi - long - range correlations in mrj packings arise from the competing effects of strict jamming and maximal randomness .
incompressibility of the structure from strict jamming implies that particles should be correlated over several characteristic length scales ; indeed , along the equilibrium branch of the binary hard - disk phase diagram , one expects that correlations become fully long - ranged at the close - packed density , corresponding to phase - separated lattice structures @xcite . however , maximal randomness interferes with this long - range order , resulting in the apparent @xmath1 asymptotic scaling in the pair correlation function . for our binary packings , this
scaling is encoded in the partial pair correlations of the structure , which , after appropriate weighting with the shape information of the particles , induces hyperuniformity .
we emphasize that a complete explanation for the appearance of the linear small - wavenumber region of the spectral density is intractable because the problem is inherently non - local due to hyperuniformity and the presence of qlr correlations .
furthermore , it has recently been established that decreasing the exponential scaling of the small - wavenumber is associated with greater disorder within a many - particle system , potentially even inducing clustering among the particles @xcite , an effect clearly inconsistent with strict jamming .
more generally , hyperuniformity is associated with an effective interparticle repulsion that attempts to evenly distribute the particles throughout space , and the length scale of this repulsion increases with increasing exponential scaling of the small - wavenumber region of the spectral density @xcite .
the conditions of saturation and impenetrability in an mrj packing induce structural order that apparently competes with the constraint of maximal randomness to minimize the scaling of the small - wavenumber region of the spectral density to its smallest integer value , an effect that we argue is physically tied to the void - space distribution of the system .
, as a function of the radii of the particles in the packing .
a uniform decrease in the particle radii defines a basin of hyperuniform systems.,scaledwidth=65.0% ] to motivate our discussion of the void space and its fundamental importance to polydisperse mrj packings , we mention a peculiar property of hyperuniformity in these packings .
namely , one can maintain hyperuniformity in polydisperse mrj hard - particle systems even upon shrinking the particles at a fixed size ratio so long as one does not affect the underlying statistical distribution of the point process ; this effect is illustrated in figure [ figninea ] .
to understand this behavior , we note that the leading - order term governing the expansion of the fourier transform of the particle indicator function @xmath134 is the volume of the particle @xmath147 , meaning that for small wavenumbers we have ( in the case of the binary packings ) @xmath148 which , upon rescaling @xmath149 for @xmath150 , suggests @xmath151 therefore , hyperuniformity of the mrj binary packing is not lost when performing this scaling operation .
physically , the small - wavenumber region of the spectral density effectively homogenizes the medium due to the coupling between the wavenumber and the particle radius in the indicator function @xmath134 , meaning that this region is not affected by changes at the boundaries of the particles so long as the underlying statistics of the point pattern generating the medium remain constant .
this observation suggests that a saturated and strictly jammed sphere packing extends naturally to an uncountably infinite family of hyperuniform heterogeneous media related by an appropriate scaling parameter .
one can map this behavior directly using the result , which indicates that the spectral density at small wavenumbers defines a quadratic form in the particle radii when the underlying point pattern of the disk centers is held fixed ; the corresponding curve is plotted in figure [ qformfig ] .
furthermore , this scaling has the effect of deforming the void space surrounding the particles in a _ uniform _ manner that preserves the regularity of the pore distribution even upon breaking both the jamming and saturation constraints .
the medium therefore is able to retain information from the strictly jammed configuration even upon relaxing the sizes of the inclusions .
additionally , it follows from the quadratic form that _ non - uniform _ changes in the particle radii inherently break the hyperuniformity of the packing . .
the parameter @xmath152 is the scaling factor.,scaledwidth=50.0% ] this behavior has important implications for mrj packings .
first , this effect is geometric in origin and depends on the explicit inclusion of particle shape information to appropriately balance the partial structure factors of the packing .
this observation immediately implies that `` point '' information contained in the particle centers is not sufficient to describe the system .
second , the presence of interparticle contacts is not essential for the onset of hyperuniformity and qlr correlations .
this subtle point , though easy to appreciate mathematically , is highly nontrivial from a physical perspective .
it implies that on the _ global _ scale of the microstructure , the space is regularized in such a way that hyperuniformity is preserved upon making uniform scaling deformations of the particle phase .
importantly , since the spectral density is a descriptor of both the particle and the void phases , this homogenization is invariant to the choice of reference phase .
note in particular that upon breaking the contact network of the mrj packing , the void space can become _ connected _ throughout the space , implying that the spatial statistics of the microstructure must account for correlations over large length scales .
it is for these reasons that we turn our attention to the void - space distribution in the following sections .
note that our scaling analysis does not apply if the particle radii are uniformly increased since breaking the impenetrability constraint implies that higher - order microstructural information is necessary to characterize the medium , and is no longer valid . additionally ,
if the particles are shrunk to the point that the small particle radius vanishes , then the packing will no longer be hyperuniform since important shape information about the system has been lost . in terms of the void space , the distribution of pore sizes will be skewed toward higher values due to the sudden appearance of `` holes '' in the microstructure , corresponding to the lost small particles , thereby de - regularizing the microstructure and breaking hyperuniformity .
it is clear from the discussion above that the void phase surrounding the disk inclusions of the binary mrj packings plays a significant role in characterizing local fluctuations in the medium .
indeed , we argue that the conditions of strict jamming and saturation place strong constraints on the distribution of the pore sizes , effectively regularizing the local structure around each disk such that the system is hyperuniform with respect to local - volume - fraction fluctuations even though the disk centers do not consitute a hyperuniform point pattern . to support our arguments concerning the void space , we have quantified the size of the available void space using the so - called _ complementary pore - size cumulative distribution function _
@xcite , which represents the fraction of the void space external to the inclusion phase with a pore radius larger than @xmath154 .
equivalently , if we define @xmath155 as the probability density that a randomly chosen point in the void space lies within @xmath154 and @xmath156 from the _ nearest _ point on the void - inclusion interface , then : @xmath157 figure [ porefig ] shows the cumulative pore - size distribution function for the binary mrj packings along with the corresponding results for the integer and binary integer packings and a system of equilibrium monodisperse hard disks at volume fraction @xmath158 .
the equilibrium hard disk packing is neither saturated nor jammed ; it is also known that this system is not hyperuniform @xcite . as a result , it possesses a broad distribution of pore sizes , and the probability of finding large pores ( i.e. , on the order of a particle diameter ) is nonvanishing . in contrast , both the integer packing and its binary variant possess narrow pore - size distributions with compact supports , which result from the close - packed nature of the packings .
interestingly , figure [ porefig ] indicates that the binary mrj packing also possesses a narrow pore - size distribution that essentially falls between the square and binary square packings .
this observation suggests that the void space is highly constrained by the condition of strict jamming and almost regular in its distribution . for the binary mrj packing , the square lattice @xmath159 , a saturated binary variant of @xmath159 , and a system of equilibrium hard spheres with volume fraction @xmath158.,scaledwidth=50.0% ] one can directly compute the moments @xmath160 of the pore size density from the cumulative distribution @xmath153 using the relation @xcite : @xmath161 lower - order moments of the pore - size density arise in bounds on the mean survival times and principal relaxation times of heterogeneous materials and are thus important descriptors of microstructures @xcite ( see section v.b below ) . table [ tableone ] provides the first moments and standard deviations of the pore size for the systems mentioned in figure [ porefig ] .
the average pore sizes for the hyperuniform systems are significantly smaller than for the equilibrium hard disk packing . in fact
, the pore size distribution of the binary mrj packing appears to be more localized about its mean than even the integer packing .
we conclude from this information that the available void space in the binary mrj disk packing is sufficiently restricted such that local volume fraction fluctuations decay faster than the volume of an observation window , thereby reflecting the underlying regularity of the pore space .
.first moments and standard deviations of the pore size for distributions of hard disks . [ cols="^,^,^",options="header " , ] it is interesting to note that the series lower bound provides a better estimate of the mean pore size than the poisson upper bound .
this observation is reasonable since the compact support of the lower bound more closely matches the behavior of the actual pore - size distribution , which also has compact support due to the saturation of the packing .
however , it is unclear if the constraint of saturation can be relaxed while still maintaining hyperuniformity in the medium .
one can conjecture that so long as the probability of finding a large pore decays faster than the corresponding behavior for a poisson pattern , then hyperuniformity holds .
although our results provide a quantitative basis for understanding the appearance of hyperuniformity in polydisperse mrj packings , a complete explanation for the linear scaling of the small - wavenumber region of the spectral density is still not apparent .
again , we emphasize that this problem is inherently non - local , and its solution must account for the presence of qlr correlations within the packing .
such correlations are difficult to discern with the distribution @xmath153 that we have presented here , which is essentially a one - point descriptor of heterogeneous media @xcite .
in the following section , we characterize the allowable @xmath3-particle loops within a binary mrj packing .
this analysis suggests that the variance in pore sizes is bounded by the strict jamming of the packings , supporting our argument these constraints are sufficient to induce hyperuniformity .
the preceding discussion shows how the void space is a _
fundamental _ descriptor of a two - phase random heterogeneous medium .
as we have seen , fluctuations in the local volume fraction contain information about the void space and therefore provide a more complete picture of hyperuniformity in heterogeneous media . here
we quantitatively characterize the voids in mrj binary disk packings and argue that strict jamming restricts the variance of pore sizes .
@xmath3-particle void _ is associated with a loop of @xmath3 contacting disks ( an _ @xmath3-particle loop _ ) in which each disk only contacts two neighbors .
these loops are defined topologically , meaning that their identification is invariant to local shears of the particle contacts .
the smallest loops contain three mutually contacting particles .
the constraint of saturation places an upper bound on the number of particles in a loop , and the largest loop we find in the mrj binary disk packings contains 6 particles .
the number of @xmath3-particle loops in a packing decreases rapidly as @xmath3 increases .
the area of the void associated with a loop can be rigorously computed by substracting the area of particles falling into the polygon constructed by connecting the centers of the disks in the loop .
we note that , except for the 3-particle loop , the polygons associated with other @xmath3-loops ( @xmath162 ) can be continuously deformed while maintaining the contacts between the particles .
such deformations change the area of the voids associated with the loop .
the rigidity of mrj packings also strongly constrains the number of particles in a loop . in the mrj packings of binary disks ,
the majority of the voids involve 3-particle loops , which form the `` backbone '' of the network .
certain 4-particle loops can be observed at the effective grain boundaries between disks with different sizes .
a large portion of these 4-particle loops are very distorted with a void area almost equal to that associated with two 3-particle loops .
loops with more particles are rare in the packings , and the observed ones are all strongly distorted with void areas almost equal to those associated with 3- or 4-particle loops .
therefore , we only discuss the voids associated with 3- and 4-particle loops here , focusing on the maximum possible void areas associated with these loops . there are four distinct 3-particle loops in the binary disk packings as shown in figure [ fig3loop ] . defining the radii of small and large disks
be @xmath163 and @xmath164 , respectively , the @xmath3-particle void areas @xmath165 , normalized by the volume of a small particle , are @xmath166 , @xmath167 , @xmath168 and @xmath169 for a size ratio @xmath170 .
it can be seen clearly that the void areas constitute a small fraction of the small particle area .
since these 3-particle voids dominate the packing , it is reasonable to conclude that the regularity of these local clusters enforces hyperuniformity on the medium despite the nonuniform distribution of the sphere centers .
there are total of six distinct types of 4-particle loops ( see figure [ fig4loop ] ) in the binary disk packings , each associated with a void that can possess a spectrum of shapes and sizes by distorting the quadrangle formed by the centers of the disks .
although most of the void areas associated with these highly distorted 4-particle loops are almost equal to the area associated with two 3-particle loops , there are still relatively large voids .
the maximum normalized void areas are given by @xmath171 , @xmath172 , @xmath173 , @xmath174 , @xmath175 and @xmath176 .
these values are relatively large but more tightly distributed around the mean when compared to the normalized areas for 3-particle voids . as a result
, they can be considered as an estimate of the upper bound on the degree of local inhomogeneity that can be consistent with hyperuniformity of the packings .
we have provided a detailed study of local - volume - fraction fluctuations in mrj packings of polydisperse hard disks and have shown that these systems are hyperuniform with quasi - long - range correlations .
our results strongly suggest that these qlr correlations are a signature of mrj hard - particle packings , in contrast to previous misconceptions in the literature @xcite .
although it is true that the structure factors for these systems do not vanish at small wavenumbers except in the monodisperse limit , we have shown that the more appropriate structural descriptor of mrj packings is the spectral density , which accounts appropriately for the shape information of the particles .
our work therefore generalizes the torquato - stillinger conjecture for hyperuniform point patterns by suggesting that all saturated , strictly - jammed sphere packings are hyperuniform with respect to local - volume - fraction fluctuations .
furthermore , mrj sphere packings are expected to exhibit quasi - long - range pair correlations scaling as @xmath1 in @xmath2 euclidean dimensions .
importantly , this generalization contains previously - published results for monodisperse mrj packings @xcite as the special case where the particle shape information can be rigorously separated from the `` point '' information of the particle centers .
based on the observations that local - volume - fraction fluctuations are invariant to the choice of reference phase and that one can maintain hyperuniformity in mrj polydisperse packings under a uniform scaling deformation of the particles , we have argued that the onset of hyperuniformity and quasi - long - range correlations results from the homogenization of the void space external to the particles .
the conditions of saturation and strict jamming limit the sizes and shapes of the local voids , which are completely determined by the contact network of particles . using a local analysis of the voids and rigorous bounds on the pore - size distribution
, we have argued that the void space is highly constrained by strict jamming , thereby suppressing infinite - wavelength local - volume - fraction fluctuations .
furthermore , we suggest that the presence of quasi - long - range correlations reflects the inherent structural correlations from the contact network between void shapes . specifically , saturation and strict jamming of the packings compete with the maximal randomness of the particle distributions to drive the small - wavenumber region of the spectral density to its smallest integer value .
although our work has addressed important problems related to the structural properties of mrj hard - particle packings , a number of unanswered questions remain .
first , a rigorous foundation for the observed linear scaling in the small - wavenumber region of the spectral density is still missing .
this problem is immensely difficult to handle theoretically because the problem is inherently long - ranged , and any local analysis of the mrj structure will therefore be unable to account correctly for this behavior @xcite .
second , although we have considered only polydisperse mrj packings of @xmath2-dimensional spheres , our results are expected to hold more generally for strictly jammed packings of hard particles of arbitrary geometry .
indeed , our arguments concerning the void space of an mrj packing are easily extended to include these more general cases , and in a companion paper we will provide direct evidence that mrj packings of hard ellipses and superdisks @xcite are also hyperuniform with signature quasi - long - range correlations indicated by linear scaling of the small - wavenumber region of the spectral density . these results will suggest a remarkably strong extension of the torquato - stillinger conjecture , namely that all maximally random strictly jammed saturated packings of hard particles , including those with size- and shape - distributions , are hyperuniform with universal quasi - long - range correlations [ via the two - point probability function @xmath177 scaling asymptotically as @xmath1 .
we mention that recent , independent work , appearing in preprint form shortly after our own manuscript was posted , @xcite has shown that hyperuniformity and vanishing infinite - wavelength local - volume - fraction fluctuations in saturated mrj packings of polydisperse disks are consistent with certain fluctuation - response relations involving a generalized `` compressibility . ''
however , this work does not address the appearance of quasi - long - range pair correlations in mrj hard - particle packings and does not consider mrj packings of nonspherical particles as in our companion paper @xcite .
our arguments suggest in particular that certain quantum many - body systems and cosmological structures with the same linear small - wavenumber scaling in the structure factor @xcite are statistically `` rigid '' in the sense that their microstructures are effectively homogeneous over large length scales with vanishing infinite - wavelength number density fluctuations .
these unique features are inherently linked to the _ structural _ properties of the system , independent of the physical model itself .
finally , we note that the systems examined in this paper are constrained to be both saturated and strictly jammed .
saturation of the packings is responsible for enforcing compact support in the pore - size distribution function and therefore plays an important role in regularizing the void space surrounding the jammed disks .
however , it is not clear if saturation is a necessary condition to ensure hyperuniformity in strictly jammed heterogeneous media ; this situation corresponds to a weakening of the torquato - stillinger conjecture .
namely , what conditions must a strictly jammed but unsaturated packing of hard spheres meet in order to be hyperuniform ?
we note that the event - driven molecular dynamics algorithm used here to generated the binary mrj disk packings inherently precludes the presence of arbitrarily large `` holes '' within the packing .
however , it does produce a small concentration ( @xmath1782.5% ) of `` rattler '' particles , which are particles that are free to move within some small caged region of the packing .
removal of such particles is known to break hyperuniformity of the medium even though the strict jamming of the surrounding structure remains @xcite ; therefore , strict jamming alone is not sufficient to induce hyperuniformity if large holes are `` common enough '' in the statistics of the microstructure .
this scenario corresponds to skewing the pore - size distribution and thereby de - regularizing the void space .
it is indeed possible to construct strictly jammed packings of two - dimensional disks with a hole of arbitrarily large size by starting with a ring of particles encompassing a large pore ; this ring can be jammed by surrounding it with a close - packed collection of particles that approximates an impenetrable `` wall '' ( see figure [ hole ] ) .
one is then free to construct any strictly jammed system of particles outside the hole @xcite .
if such holes are sufficiently rare , then it is possible that the system may still be hyperuniform since the pores do not become very large on average .
the converse statement is not true , meaning that hyperuniformity does not imply strict jamming .
the square lattice in two dimensions is one example of a hyperuniform packing that is not strictly jammed . a _ bravais lattice _
is defined by the integer linear combinations of a set of @xmath2 basis vectors in @xmath4 . by taking the union of a bravais lattice with its translate(s ) by one or more vectors
, one can also form periodic non - bravais lattices , also known as lattices with a basis in the physics literature .
unless otherwise stated , in this paper we use the term _ lattice _ to refer exclusively to a bravais lattice .
although we only consider packings in 2d for this study , our results are expected to apply in higher dimensions as well .
we have already seen that our methodology contains known results for 3d monodisperse mrj sphere packings as a special case , and the arguments we make in this section concerning the constrained void space extend directly to higher - dimensional packings .
it is not clear if this analogy can be extended to higher dimensions , where it is intimately related to the problem of finding optimal spherical codes ; see : j. h. conway and n. j. a. sloane , _ sphere packings , lattices and groups _ ( springer , new york , 1999 ) , 3rd ed .
|
hyperuniform many - particle distributions possess a local number variance that grows more slowly than the volume of an observation window , implying that the local density is effectively homogeneous beyond a few characteristic length scales .
previous work on maximally random strictly jammed sphere packings in three dimensions has shown that these systems are hyperuniform and possess unusual quasi - long - range pair correlations decaying as @xmath0 , resulting in anomalous logarithmic growth in the number variance .
however , recent work on maximally random jammed sphere packings with a size distribution has suggested that such quasi - long - range correlations and hyperuniformity are not universal among jammed hard - particle systems . in this paper
we show that such systems are indeed hyperuniform with signature quasi - long - range correlations by characterizing the more general local - volume - fraction fluctuations .
we argue that the regularity of the void space induced by the constraints of saturation and strict jamming overcomes the local inhomogeneity of the disk centers to induce hyperuniformity in the medium with a linear small - wavenumber nonanalytic behavior in the spectral density , resulting in quasi - long - range spatial correlations scaling with @xmath1 in @xmath2 euclidean space dimensions .
a numerical and analytical analysis of the pore - size distribution for a binary mrj system in addition to a local characterization of the @xmath3-particle loops governing the void space surrounding the inclusions is presented in support of our argument .
this paper is the first part of a series of two papers considering the relationships among hyperuniformity , jamming , and regularity of the void space in hard - particle packings .
|
for several decades , the standard model has been remarkably successful in explaining and predicting experimental data .
however , the mechanism of electroweak symmetry breaking is still not known .
the most popular mechanism to induce spontaneous symmetry breaking of a gauge theory , resulting in the gauge bosons and fermions acquiring masses , is the higgs mechanism @xcite , which predicts the existence of a higgs particle with unknown mass .
the current direct search limit @xcite for standard model higgs boson at lep-2 is @xmath4 gev / c@xmath1 at 95% c.l .
lep-2 will continue their run until the end of year 2000 , which will allow them to either exclude the higgs mass up to 110 gev / c@xmath1 or have a 5 @xmath5 higgs discovery for higgs mass below 106 gev / c@xmath1 .
the indirect search via electroweak precision tests through radiative correction yields @xmath6 gev / c@xmath1 at 95% c.l . at the tevatron ,
one of the higgs production processes more likely to be observed is an associated production @xmath7 for @xmath8 gev / c@xmath1 , where v = w , z and @xmath9 . for the higgs mass above 130 gev / c@xmath1
, different search strategies need to be developed since the dominant decay mode is no longer @xmath10 , but @xmath11 @xcite
. the experimental signature considered is @xmath13 with @xmath14 or @xmath15 , and @xmath9 , giving final states with one high-@xmath16 lepton , large missing transverse energy ( @xmath17 ) due to the undetected neutrino and two @xmath18 jets .
the ability to tag @xmath18 jets with high efficiency and a low mistag rate is vital for searching for the decay of @xmath9 .
cdf uses the secondary vertex ( secvtx ) and soft - lepton ( slt ) @xmath18-tagging algorithms developed for the top quark discovery , while d0 uses a soft - muon @xmath18-tagging only .
both cdf and d0 select the @xmath18-tagged @xmath19 2 jet events since it is expected to contain most of the signal , while @xmath18-tagged @xmath20 jet events are dominated by @xmath21 decays . in a data sample of @xmath22 pb@xmath0 ,
cdf observed 36 events with single svx @xmath18-tag and 6 events with two @xmath18-tagged jets ( svx - slt and svx - svx ) @xcite .
the expected backgrounds are @xmath23 for single and @xmath24 for double tagged events , which are predominately from @xmath25 , @xmath26 , mistags and @xmath21 decays .
the probability that the background fluctuated upward to the number of observed events is found to correspond to one standard deviation .
the dijet mass distribution for the single and double tagged events are shown in figure [ bbmass ] , along with the background expectation .
the single - tag data show a slight excess of events at higher two - jet mass but there is no mass peak as would be expected from the two - body decay of a new particle .
likelihood fits to the mass distributions yield a 95% c.l .
limit on the production cross section times branching ratio , shown in figure [ bblimit ] , as a function of higgs mass . 2.0 truein 1.0 cm 2.0 truein 1.0 cm 2.0 truein
-.2 cm d0 also performed similar search in their 100 pb@xmath0 lepton + jets data @xcite .
they observed 27 events with at least one @xmath18-tag , which are in good agreement with the background expectation of @xmath27 events . the observed dijet mass distribution and corresponding 95% c.l .
limit on the production cross section times branching ratio are shown in figure [ bbmass ] and [ bblimit ] .
cdf also searches for higgs in the four - jets channel of @xmath29 process , where v = w , z decays into two jets @xcite .
the events are required to have at least two svx @xmath18-tagged jets and the @xmath16 of two @xmath18-jets greater than 50 gev / c@xmath1 in order to reduce the large qcd background . in a multi - jet trigger sample of @xmath30 pb@xmath0 ,
cdf observed 589 candidate events .
the invariant mass of two @xmath18-tagged jets is shown in figure [ bbmass ] , along with the main backgrounds from qcd , fake tags and @xmath21 decay .
a likelihood fit is then applied to the shape of signal plus background , with the normalization of signal and qcd free , which yields total 600 background events and consistent with zero signal events . the corresponding 95% c.l .
limits on the production cross section times branching ratio is shown in figure [ bblimit ] .
it also shows the cdf combined limit obtained from both all - hadronic and lepton + jets channels .
several studies @xcite in past have indicated that there is significant potential of observing associated production of higgs and @xmath32 boson with @xmath33 , because of large branching rations of @xmath34 .
the signature of such decay is relative clean , a resonance of two @xmath18 tagged jets , plus either large missing @xmath35 ( @xmath36 gev ) or dilepton pair mass near the @xmath32 mass , which were triggered experimentally by either @xmath37 or leptons in the event .
the @xmath17 in the background sample is predominately due to mismeasured jets resulting in the @xmath17 direction being aligned along with the jet direction while the @xmath38 in the signal sample is flat . by requiring large missing @xmath39 ( @xmath36 gev ) and @xmath40 ,
the qcd background can be reduced to minimum , which results in a higgs reach sensitivity similar to the one in @xmath41 channel .
the analysis of cdf search in run1 is in progress .
d0 has searched for @xmath42 in their run1 data .
they observed 2 events in the data , in good agreement with the expected background of @xmath43 events .
the corresponding 95% c.l .
limits on the production cross section times branching ratio is shown in figure [ bblimit ] .
2.0 truein 1.0 cm 3.0 truein -.2 cm during feb.- nov .
1998 , there was a joint cdf / d0/theory working group to study higgs searches at run - ii or beyond at the tevatron using more realistic cdf / d0 detector simulation for the signal and background estimation ( see http://fnth37.fnal.gov/higgs.html for more details ) . since the tevatron would be the only chance to discover the higgs before lhc , it would be important to establish the realistic higgs discovery potential beyond lep-2 .
great progress has been made in terms of extending the depth and breadth of previous studies on the higgs reach @xcite .
there is no single , golden discovery channel at the tevatron .
combining all the channels and data from both experiment data is crucial .
the preliminary results indicate that the tevatron should be able to exclude the higgs mass up to 130 gev / c@xmath1 at the 95% c.l . with 2 fb@xmath0 data , or have a 5 @xmath5 discovery of the higgs mass up to 130 gev / c@xmath1 with 30 fb@xmath0 data @xcite .
a precise measurement of the top quark mass is an important ingredient in testing the consistency of the standard model with experimental data . in addition ,
precise @xmath44 and top mass measurements can provide information on the mass of the higgs boson , which is a remnant of the mechanism that gives rise to spontaneous electroweak symmetry breaking . within the framework of the standard model the top quark decays almost exclusively into a real w boson and a b quark .
the observed event topology is then determined by the decay modes of the two @xmath44 bosons , which can be classified into three decay channels .
the dilepton channel has about 5% of the cases , with both @xmath44 bosons decay to @xmath45 or @xmath15 .
the lepton + jets channel has 30% of the cases , one @xmath44 boson decays to @xmath45 or @xmath15 , and the other to a @xmath46 pair .
the all - hadronic channel has @xmath47 of the cases , involving the hadronic decay of both @xmath44 bosons . .
the observed number of events and expected backgrounds for the top decay by the cdf and d0 experiments . [
cols="^,^,^,^,^",options="header " , ] table i. shows that cdf and d0 observe a clear excess of events over the expected background in the dilepton , lepton + jets , and all - hadronic channels .
the two experiments taking slightly different approaches in defining their events samples , with cdf taking advantage of their silicon vertex detector ( svx ) and d0 making greater use of kinematic variables to reduce backgrounds . the present sample of top candidates are consistent with standard model decays .
one of cdf and d0 physics goals in run1 is to determine the top mass as accurately as possible using many different channels and techniques . as a result ,
the uncertainty on the top quark mass has improved from @xmath48 gev / c@xmath1@xcite , first published in 1994 to the present @xmath49 gev / c@xmath1 .
the advantage of measuring a top quark mass in the lepton + jets channel is its relatively larger branching ratio and the ability to full reconstruct the top mass on an event - to - event basis .
both cdf and d0 select events containing a high @xmath39 ( @xmath16 ) single isolated electron ( muon ) in the central region , large missing transverse energy and at least four jets .
measurement of the top quark mass begins by fitting each event in the sample to the hypotheses of @xmath21 production followed by decay in the lepton + jets channel ( @xmath50 ) .
there are twelve distinct ways of assigning the four leading jets to the four partons @xmath18 , @xmath51 , @xmath52 and @xmath53 .
in addition , there is a quadratic ambiguity in the determination of the longitudinal component of the neutrino momentum .
this yields up to twenty - four different configurations for reconstructing an event according to the @xmath21 hypothesis with no @xmath18-tag , 12 configurations for events with a single @xmath18-tag , and 4 configurations for events with double @xmath18-tags .
both cdf and d0 choose the jet configuration with the lowest @xmath54 .
the precision of the top quark mass measurement is expected to increase with the number of observed events , the signal - over - background ratio , and the narrowness of the reconstructed - mass distribution . in cdf lepton
+ jets mass analysis @xcite , monte carlo studies show that an optimum way to partition the sample consists of subdividing the events into the four statistically independent subsamples : svx single @xmath18-tag , svx double @xmath18-tag , slt @xmath18-tag and untagged events .
the reconstructed - mass distributions of the four subsample is plotted in figure [ cdflj ] , from which cdf measures @xmath55 gev / c@xmath1 .
the d0 lepton + jets mass analysis uses four kinematic variables weakly correlated with top mass to separate the top signal from the background processes @xcite .
they were combined into two different multivariate discriminant : a weighted likelihood that minimizes correlation with the reconstructed mass ( lb or low bias " ) and a neural network output ( nn ) .
the data are binned according to the discriminant value to create top - rich and background - rich sub - samples .
the subsamples are simultaneously fit to obtain the top quark mass @xmath56 gev / c@xmath1 , shown in figure [ cdflj ] .
2.5 truein 1.0 cm 2.5 truein cdf recently reported an improved measurement of the top quark mass using dilepton events @xcite originating predominantly from @xmath57 , where @xmath58 or @xmath59 .
this measurement supersedes their previously reported result in the dilepton channel @xcite , which was obtained by comparing data with monte carlo simulation of @xmath21 events for two kinematic variables , the @xmath18-jet energies and the invariant masses of the lepton and @xmath18-jet systems .
since the dilepton system is under - constrained due to the two missing neutrinos in the final state , the cdf and d0 dilepton mass analyses @xcite proceed by hypothesizing a top mass ( @xmath60 ) and solving event kinematics up to four - fold ambiguity for each of the two lepton - jet parings .
two different weighting techniques have been developed .
* neutrino weighting technique assigns a weight to each solution by comparing the predicted and measured missing transverse energy ( cdf and d0 ) @xcite . *
matrix element weighting technique assigns a weight to each solution by the parton distribution functions and the matrix element for @xmath44 boson decay ( d0 ) @xcite . for both weighting methods ,
the weights are summed over each lepton - jet paring , and over the detector resolution for jets and leptons by sampling ( @xmath61 fluctuating ) the measured quantities many times according to their detector resolutions .
the resulting weight functions for each cdf and d0 dilepton events , normalized to unity , are shown in figure [ dilw ] .
truein 1.0 cm 2 .
-.2 cm cdf assign each event a mass by averaging the two mass values corresponding to the weight closest to and greater than the half maximum weight on either side .
figure [ cdfdilm ] shows the distribution of these masses , together with the monte carlo expectation for background and fitted top , from which cdf determines a top quark mass value of @xmath62 gev / c@xmath1 .
2.0 truein 1.0 cm 2.0 truein -.2
cm d0 accounts for the shape of the weight function by integrating the weights in five equally spaced mass bins and forming a four - dimensional vector from the normalized weights .
the top quark mass is extracted by comparing the distribution of observed events in this four - dimensional vector space with the predicted densities for top signal and background .
the results of these fits give a top quark mass value of @xmath63 gev / c@xmath1 . in this analysis @xcite cdf
selects @xmath21 events in which both @xmath44 bosons decay into quark - antiquark pairs , leading to an all hadronic final state .
the study of this channel , with a branching ratio of about 4/9 , complements the leptonic modes , and the mass measurement takes advantage of fully reconstructed final state , but suffers a very large qcd multijet background . to reduce this background , events with at least one identified svx @xmath18-jet are required to pass strict kinematic criteria that favor @xmath21 production and decay . the data sample consist of 136 events , of which @xmath64 events are expected to come from background .
events are then reconstructed to the @xmath65 hypothesis , where both @xmath44 bosons decay into a quark pair , with each quark associated to one of the six highest @xmath39 jets .
all the combinations are tried and the combination with lowest @xmath66 is chosen .
the reconstructed 3-jet mass distribution is shown in figure [ cdfdilm ] , from which cdf measures a top quark mass of @xmath67 gev / c@xmath1 .
the overall systematic error has been revised from 12.0 to 5.7 gev / c@xmath1 @xcite .
the individual results from both cdf and d0 are combined according to a procedure developed by a joint cdf - d0 working group .
the combined top mass from the tevatron is @xmath68 gev / c@xmath1 .
the good agreement among the measurements is reflected by a @xmath54 probability of 75% of the mass average . combining the statistical and systematic errors in quadrature
, we find @xmath69 gev / c@xmath1 , which gives the best measured quark mass ( @xmath70 ) . in run - ii with 2 or more fb@xmath0 of integrated luminosity and the increased capabilities of the upgraded cdf and d0 detectors
, we expect more than 1000 single tagged and about 600 double tagged @xmath21 events for each experiment .
it will allow us to measure @xmath60 down to statistical error of 0.5 gev / c@xmath1 by scaling the present results .
the dominant systematic errors are expected to be the uncertainties in modeling gluon radiation and the jet energy scale of the detectors , which hopefully can be reduced to 2 gev / c@xmath1 by studying @xmath71 in the double tagged @xmath21 events and 32k @xmath72 events from secondary vertex trigger .
the author is grateful to all members of the cdf and d0 top and exotic physics groups for providing material in this paper .
especially , i would like to thank for : e. barberis , j. conway , r. demina , l. galtieri , j. lys , m. shapiro , f. stichelbaut and d. wood for their useful comments .
|
a summary of the present standard model higgs search and measurement of top quark mass at the tevatron are presented .
the sensitivity of the present higgs search at the tevatron is limited by statistics to a cross section approximately two orders of magnitude higher than the predicted cross section for standard model higgs production . with 30 fb@xmath0 of integrated luminosity , the tevatron offers an unique potential discovery window for the standard model higgs mass up to 130 gev / c@xmath1 before lhc era .
the study of top at the tevatron has moved from discovery phase to one of characterizing its properties .
the combined result of top quark mass is @xmath2 gev / c@xmath1 ( @xmath3 ) .
|
wide varieties of magnetic ground structures are manifested in @xmath0ni@xmath3b@xmath1c series ( see e.g. refs.@xcite and tab.i ) : members with @xmath18ho , dy , nd , pr adopt commensurate antiferromagnetic ( af ) arrangements while those with @xmath18tm , er , tb , gd , assume squared - up spin - density wave ( sdw ) states .
these ground structures are stabilized by a fine balance of exchange , crystalline electric field ( cef ) , and magnetoelastic forces . under magnetic field or temperature variation ,
most of these structures undergo a cascade of phase transformations , yielding rich varieties of field - temperature ( @xmath19 ) phase diagrams ( see e.g. refs.@xcite ) .
interestingly , most of the zero - field magnetic states of @xmath0=tm , er , ho , dy coexist with superconductivity , presenting model compounds wherein the interplay between superconductivity and magnetism can be investigated .
such investigations revealed that the superconductivity , though much influenced by , has a very weak influence on the prevailing magnetic order:@xcite the energy gain due to the onset of magnetic order dominates by two order of magnitudes over that due to the onset of superconductivity .
the zero - field part of the @xmath19 phase diagram of honi@xmath1b@xmath1c is particularly interesting : superconductivity sets - in at @xmath20 k. just below this point , an incommensurate spiral state , @xmath21.92@xmath22 , develops ( see e.g. refs.@xcite ) .
furthermore , at @xmath236.3 k , an additional modulated state with @xmath240.55@xmath25 emerges and around 5 k a deep minimum in @xmath26 develops.@xcite at @xmath275 k , an orthorhombic lattice distortion @xcite sets - in and , concomitantly , both the spiral and the a - axis modulated states are replaced by a commensurate af structure which coexists with superconductivity down to the lowest measurable temperatures .
generalized susceptibility calculation @xcite related these modulated state to maxima in the exchange coupling transform,@xmath28 . remarkably , the magnetic ground structures of the heavy and magnetic @xmath0ni@xmath1b@xmath1c compounds ( see tab.i ) are particularly simple : in spite of the manifestation of an orthorhombic distortion and a liquid - helium - temperature magnetic modulation , their ground structures are either an equal - amplitude , af - type squared - up state ( as in @xmath18tm , er , tb , gd ) or an equal - amplitude , colinear , commensurate af state ( as in @xmath18ho , dy ) . then , it is of interest to investigate whether the low - temperature thermodynamics of these @xmath0ni@xmath1b@xmath1c can be described in terms of small - amplitude spin - wave excitations and , in addition , to elucidate the character and dimensionality of these excitations .
these magnetic excitations can be probed by various techniques , among which is the magnetic specific heat .
we carried out extensive zero - field specific heat measurements on five single crystals @xmath0ni@xmath1b@xmath1c
( @xmath0= er , ho , dy , tb , gd ) covering at least the range @xmath29 k@xmath30 these specific heats , together with that of tmni@xmath1b@xmath1c ( ref.@xcite ) , reveal a diversified and wide varieties of thermal evolutions .
none the less , based on a phenomenological model , all of the specific heat curves can be systematized in terms of only two parameters , namely an effective exchange coupling and a magnetic anisotropy interaction .
the format of this paper is as follows : in sec.ii , we derive an approximate , but of wide applicability , expression for the magnon specific heat . experimental techniques and procedures
are described in sec.iii .
results and their analysis are described in sec.iv and discussed in sec.v .
the magnetic structures of @xmath0ni@xmath1b@xmath1c ( @xmath0=tm - gd ) can be visualized as magnetic layers that are stacked along the c - axis.@xcite the most dominant interactions are the rkky and anisotropic couplings .
the former can be approximated by effective isotropic couplings while the latter ( a combination of dominant cef and weaker dipolar and anisotropic exchange forces ) by an easy - axis anisotropy field @xmath31:@xcite this @xmath31 representation is convenient for spin - wave calculation and is valid for the low - temperature orthorhombic - distorted equal - amplitude collinear af or squared - up phases of borocarbides ( see sec.v ) .
considering the above mentioned magnetic arrangement and the crystal symmetry , the magnetic couplings can be conveniently divided into two classes : @xmath32 that couples moments @xmath33 and @xmath34 within the same layer ( denoted as @xmath35 or @xmath36 ) and @xmath37 that couples moments from different layers . then at zero external field , the following hamiltonian is expected to capture most of their low - temperature properties : @xmath38 the first and second sum represent , respectively , interactions within the same layer and among different layers .
the last two terms represent the anisotropic interactions . by standard spin - wave analysis
, we obtained the following dispersion relation : @xmath39 ^{2}-\left [ sj_{\vert}(k)\right ] ^{2 } } \label{dispersion}%\ ] ] where @xmath40 @xmath41 ( @xmath42 ) represents the fourier - transform parallel ( perpendicular ) to the c - axis .
the energy gap , @xmath43 is : @xmath44 which defines an af resonance frequency similar to the uniform mode of ordinary afs.@xcite evidently , ( i ) @xmath4 is zero whenever there is no anisotropy and ( ii ) @xmath4 does not depend on the type nor the strength of the intralayer coupling . for evaluating the magnon specific heat ( or other thermodynamical quantities ) , an explicit expression of @xmath45
is required . in the absence of such an expression and for low - temerpature range , it is a common practice to assume a long - wave limit . here , we restricted the expansion of @xmath45 to the nearest neighbors only , leading to : @xmath46 where @xmath47 ( @xmath48 for weak orthorhombic distortion ) and @xmath49 are functions of the exchange couplings and geometrical factors ( @xmath50 and @xmath51 are unit - cell parameters):@xmath52 then , the zero - field magnon specific heat is ( rewritten so as to conform with the notation of ref.@xcite ) : @xmath53 \label{cm - bessel}%\ ] ] where @xmath54 represents the modified bessel function and @xmath55{3^{\frac{3}{2}}2(c_{x}% c_{y}c_{z})^{\frac{1}{2}}/(a^{2}c ) } \label{theta}%\ ] ] is a characteristic temperature , based on which @xmath56 can be defined as being an effective exchange interaction that couples the magnetic moment to its @xmath57 nearest neighbors .
notice that both eq.[cm - bessel ] and eq.[theta ] account as well for the weakly orthrohmic - distorted state
. for @xmath58 , eq.[cm - bessel ] reduces to the exponential form : @xmath59 while for isotropic compounds or @xmath60 , it reduces to the high - temperature limit : @xcite @xmath61 in this case , eq.[theta ] highlight the useful definition of @xmath62 .
a long - wave dispersion relation for an isotropic quasi-__2__@xmath63 case can be derived from eq.[dispersion - quadratic ] , if we set @xmath64 ( @xmath65 ) and @xmath66 : @xmath67 then , to lower order in @xmath68 , one obtains @xmath69 which reproduces the leading linear - in-@xmath70 term in the expression reported by movshovich et al @xcite who ( starting from a quadratic dispersion relation and including correction for the 2d and the magnon - magnon interaction ) obtained for the range@xmath71:@xmath72 it is worth remarking that eqs.[h]-[dispersion - quadratic],[cm - bessel ] are of a more wide applicability than our above analysis might have suggested . furthermore , depending on only @xmath5 and @xmath4 , a variety of expressions for the magnetic specific heat can be derived ; each admitting different limits for its thermal evolution depending on the relations among @xmath70 and @xmath4 ( compare eq.[cm - bessel ] with the limit eqs.[cm - af - lowt],[cm - af - hight],[cm-2d ] ) .
based on such a scheme , one is capable of rationalizing the vast varieties of the low - temperature thermal evolution of thermodynamical quantities ( such as @xmath73 ) encountered in these ( and any series similar to ) borocarbides .
it is reminded that this analysis is not adequate for the description of the contribution of the modulated state since the involved ordered components do not have equal magnitudes . finally , the above zero - field treatment
is obviously not appropriate for the description of the field - induced metamagnetic phases .
single crystals of @xmath0ni@xmath1b@xmath1c ( @xmath0=er - gd ) were grown by floating zone method.@xcite physical characterization were carried out utilizing a dc magnetometer [ @xmath74@xmath7550 koe , 2 k @xmath75@xmath70@xmath7520 k ] , a mutual - induction ac - susceptometer [ @xmath76=250 - 500 hz , @xmath771 oe , 30 mk @xmath75@xmath70@xmath7520 k ] , and a dc resistivity set - up [ @xmath74@xmath7550 koe , 2 k @xmath75@xmath70@xmath7520 k ] . structural , magnetic and transport characterizations are in agreement with published results .
the temperature - dependent specific heat was measured on a semi - adiabatic calorimeter [ 80 mk @xmath75@xmath70@xmath7525 k , precision better than 4% ] .
the total specific heat curves measured above 2 k are in agreement with the reported data.@xcite however , we observed some discrepancy between the absolute values of @xmath73 of single crystal and polycrystalline samples : though both specific heats were found to be given by approximately the same functional form , the absolute values of the fit parameters ( @xmath5 and @xmath4 ) differ by as much as 40% . for each compound , the total specific heat , @xmath78 , was analyzed as a sum of an electronic @xmath79 ( @xmath80 when superconductivity is to be emphasized ) , a debye @xmath81 @xmath82 , a nuclear @xmath83 , and a magnetic contribution @xmath84 from the only magnetically active @xmath0-sublattice . at temperatures of interest , @xmath79 and @xmath81
were estimated based on our specific heat characterization @xcite of single crystal yni@xmath1b@xmath1c ( @xmath85 mj / molek@xmath86 and @xmath87 0.12 mj / molek@xmath88 ) which had been synthesized by the very same procedures as the one used for the other single crystals .
thus for each compound , @xmath89 was taken to be the same value as that of yni@xmath1b@xmath1c while @xmath90 was evaluated by normalizing the @xmath90 of yni@xmath1b@xmath1c using the approximation @xmath91(see tab.ii ) . within the superconducting region ,
@xmath80 was evaluated as 3@xmath92 . at any rate , for all the studied compounds
, @xmath73 is much larger than the sum of @xmath79 and @xmath81 .
consequently , even if @xmath79 and @xmath81 are taken as the bare values of yni@xmath1b@xmath1c , @xmath73 would not be noticeably modified , ensuring that our conclusions would not be influenced .
@xmath93 of @xmath0ni@xmath1b@xmath1c , when available , is of dominant importance only at very low temperatures and was evaluated by least - square fit using the appropriate hyperfine hamiltonian ; the obtained parameters ( shown in tab.ii ) compare favorably with those of the corresponding @xmath0-metal @xcite and @xmath0co@xmath1b@xmath1c isomorphs.@xcite
from the general feature of @xmath73 curves ( figs.1 - 3,5 - 6 ) , one distinguishes four temperature regions : ( i ) a paramagnetic region , @xmath94 , wherein @xmath73 is predominantly due to a change in the population of the cef levels , ( ii ) a critical region , @xmath95 , wherein @xmath73 is related to the critical phenomena , ( iii ) an intermediate region wherein @xmath73 reflects the magnetic character of the spiral / modulated states , and ( iv ) the low - temperature af / squared - up states , of prime interest to this work , wherein the measured @xmath73 is to be confronted with eqs.[cm - bessel],[cm-2d ] and therefrom @xmath96 and @xmath4 are to be extracted .
before we discuss the features of @xmath73 for each compound , a word of caution is in order : just as in the case of @xmath0-metals,@xcite the propagation of errors due to successive subtraction of @xmath97 , @xmath98 and @xmath93 would eventually influence the absolute value of @xmath73 . below @xmath99 , the zero - field magnetic structure @xcite is a transverse sine - modulated type with @xmath100 that changes from 0.551@xmath25 at @xmath99 to 0.550@xmath25 at @xmath101 , where a spin reorientation process sets - in . below @xmath101
, @xmath100 reverts course and increases monotonically till reaching 0.553@xmath25 at 3.5 k. @xmath73 of gdni@xmath1b@xmath1c ( fig.1 ) reveals the onset of the magnetic order at @xmath2719.5 k and the spin reorientation process at @xmath10213.5 k , in agreement with earlier studies .
@xcite the thermal evolution of @xmath73 within the amplitude - modulated state is distinctly different from that within the equal - amplitude , low - temperature state ( see below ) . as mentioned above , within the modulated region the linearized spin - wave analysis is not applicable and one should resort to the findings of schmitt and co - workers @xcite .
in particular , these authors demonstrated that @xmath103 of such a state suffers a strong reduction ( almost 1/3 ) in comparison with the value ( 20.15 j / molek ) expected for an equal - amplitude af state ( our results are in excellent agreement with this statement ) .
[ ptbh ] magnon-fig1.eps on the other hand , fig.1 shows that for temperatures below 3.5 k , @xmath73 follows faithfully eq.[cm - bessel ] with @[email protected] and @[email protected] k. the numerical value of @xmath5 ( for @xmath4 see sec.v ) is physically acceptable as can be seen from the following arguments .
first , the substitution of this @xmath5 into eq.[theta ] yields @xmath106= [email protected] k , which is in close agreement with the value reported for honi@xmath3b@xmath1c ( ref.@xcite,@xcite ) and tmni@xmath3b@xmath1c ( ref.@xcite ) .
secondly , the substitution of @xmath5 into eq.[cm - af - hight ] predicts correctly the high temperature limit , namely @xmath108 = 0.058@xmath109 j / molek ( see fig.1 ) .
thirdly , the substitution of @xmath5 into the molecular field relation:@xcite @xmath110 gives @xmath99 = [email protected] k which is in reasonable agreement with the experimentally determined value of @xmath99 .
a longitudinal sdw , accompanied by an orthorhombic distortion , sets - in at @xmath99.@xcite the magnitude of the modulation vector decreases from 0.551@xmath25 near @xmath99 to 0.545@xmath25 at 2.3 k.@xcite a weak ferromagnetic component develops below @xmath111 8 k and at lower temperature a squaring - up of the modulated state was observed.@xcite @xmath73 of single crystal tbni@xmath1b@xmath1c ( fig.2 ) shows the magnetic ordering at @xmath112 k and the wf - associated anomaly that peaks around 5.5 k. these features are in agreement with those reported by tommy et al.@xcite similar to the case of gdni@xmath1b@xmath1c , no attempt was made to analyze @xmath73 within the amplitude modulated state spanning the range @xmath113 k@xmath114 .
below 5 k , where the orthorhombic - distorted squared - up state is expected , @xmath73 follows convincingly the prediction of eq.[cm - bessel ] with @[email protected] k and @[email protected] k. [ ptbh ] magnon-fig2.eps this compound develops a commensurate af structure below @xmath99 with moments arranged in an identical manner as that of honi@xmath1b@xmath1c .
@xcite moreover , superconductivity coexists with this af order below @xmath1166 k. in contrast to other ni - based af superconducting borocarbides , dyni@xmath1b@xmath1c presents the following distinct features : ( i ) no zero - field incommensurate or modulated state was reported.@xcite however for @xmath117 2 k , anomalously large hysteresis and pronounced reentrant effects were observed for the field range 1 [email protected] koe , @xcite ( ii ) the superconductivity emerges within a well developed af order ( @xmath17 ) and that @xmath119 is extremely sensitive to nonmagnetic doping.@xcite @xmath120 ( see fig.3 ) reveals the onset of the af order at @xmath121 k. within the accuracy of our measurement , the superconducting jump at @xmath119 @xmath1226 k is too small to be resolved . on carrying out the analysis of @xmath78 into its components ( @xmath79 , @xmath123 , and @xmath124 ) , we observed an anomalous contribution peaking at @xmath125 k and having features reminiscent of a schottky - like contribution .
accordingly , it was approximated by the standard two - level relation : @xmath126 ^{2 } \label{csch}%\ ] ] where @xmath127 is the energy separation .
it was found out ( see the inset of fig.3 ) that @xmath127 @xmath128 k and that only 0.062 molar fraction was involved .
moreover , the fit was satisfactory for the high temperature tail but not so good at the lower temperature part , suggesting that a multi - level schottky contribution might be more appropriate . however , for the present discussion , the above two - level approximation is sufficient .
it is highly possible that such a contribution is due to 6% defect / impurity which is beyond the limit of detection of our x - ray structural characterization .
coincidently , anomalous hystereses effects were observed in the magnetostriction curves that were measured within the same temperature range.@xcite at any case , for @xmath129 , both @xmath130 and @xmath93 are smaller than @xmath73 ( see inset of fig.3 ) .
nevertheless , we considered @xmath131 ( see fig.3 ) .
the thermal evolution of @xmath73 is shown in a log - log plot in fig.3 and as ln(@xmath73 ) @xmath132 @xmath133 plot in fig.4 . in both figures , the comparison with eq.[cm - bessel ] ( solid line ) is also presented . evidently over a wide range of temperatures , @xmath134 follows excellently eq.[cm - bessel ] with @[email protected] and @[email protected] k. [ ptbh ] magnon-fig3.eps [ ptbh ] magnon-fig4.eps @xmath120 of single crystal honi@xmath1b@xmath1c ( shown in fig.5 ) reveals a cascade of three transitions that are usually attributed to magnetic transformations.@xcite the signature of the onset of superconductivity is too weak to be observable in our present measurements . on the other hand , for @xmath1175
k , @xmath73 follows the description of eq.[cm - bessel ] with @[email protected] and @[email protected] k. [ ptbh ] magnon-fig5.eps two intriguing features of the @xmath19 phase diagram of erni@xmath1b@xmath1c are @xmath135@xcite : the onset of the incommensurate transversely polarized sdw state with _ _ k__@xmath136=0.553@xmath137 at @xmath275.94 k , and the onset of weak ferromagnetism ( wf ) at @xmath1382.2 k. these two events ( none is able to quench superconductivity , @xmath139 k ) are well evident in @xmath73 of fig.6 .
@xmath140 , in particular , is evident as a change of slope that separates two distinct thermal evolutions:@xcite @xmath73 within the amplitude - modulated state @xmath141 and that within the squared - up state at @xmath142 . in the latter region
, @xmath143 is well described by eq.[cm - bessel ] with @[email protected] k and @[email protected] k. [ ptbh ] magnon-fig6.eps superconductivity sets - in at 11 k and , below @xmath146 k , coexists with a transversely polarized sdw state wherein the spins are pointing along the c - axis and the modulation vector is ( 0.093,0.093,0).@xcite at lower temperatures , the incommensurate sdw state squares up.@xcite the magnetic specific heat of single crystal tmni@xmath1b@xmath1c was measured by movshvich et al @xcite and was shown to follow eq.[cm-2d ] with @xmath1470.8 k , @xmath1480.2 k.
on a linearized spin - wave ( non interacting magnon gas ) approach , one is always limited to the low - temperature region of the ordered state which in the case of the borocarbides amounts to being restricted to below the liquid helium temperatures .
within that temperature range , the model describes very successfully the magnetic contribution of the studied compounds indicating that @xmath84 can be safely associated with the gapped collective excitations that propagate within the orthorhombic - distorted 3d af ( -type ) structure .
the successful applicability of the model to the commensurate collinear af ground structures of @xmath18ho , dy is understandable . to justify its applicability to the cases of @xmath18tm , er , tb , gd , it is sufficient to show that their ground structures are well squared - up and that all the moments do have equal amplitudes and orient ( or bunch ) along a specific direction .
the state of erni@xmath1b@xmath1c below @xmath140 and that of gdni@xmath1b@xmath1c below 3.5 k provide the best illustrations of the fulfillment of this requirement .
the colinear , equal - amplitude , and squared - up character of the state of erni@xmath1b@xmath1c below @xmath140 was elegantly revealed in the neutron diffractions studies of choi et al@xcite and kwano - furukawa et al @xcite .
it is remarked that the presence of weak ferromagnetism ( reflected as kink that separate oppositely oriented domains ) would hardly modify this picture since the excitation energy of the kink is much higher than that of the magnon .
let us now discuss the character of the state of gdni@xmath1b@xmath1c below 3.5 k. a recent magnetoelastic study @xcite on single crystal gdni@xmath1b@xmath1c demonstrated the presence of substantial magnetoelastic and anisotropic exchange interactions , in particular below @xmath149 .
the magnitude of the @xmath150 strain mode ( which leads to symmetry lowering from four - fold tetragonal to two - fold orthorhombic ) is very large and increases with decreasing temperature leading progressively to an orthorhombic - distorted magnetic state wherein the moments , due to entropy arguments , approach equal amplitudes .
based on the suggestion of detlefs et al @xcite , the magnetic structure below @xmath101 is either a transverse modulated state with moment orientation away from the b - axis in the bc - plane or a modified spiral - like structure which , due to the low symmetry ( 2 mm ) of the [ 100 ] direction , is likely to suffer fanning or bunching ( becoming more stronger as the temperature is decreased ) .
tomala et al @xcite , on exploring these two structural possibilities , argued that the @xmath151gd mossbauer spectra at 4.2k @xmath75@xmath149 were better fitted with a bunched spiral - like state . in this mossbauer analysis ,
the c - component was found to be almost equal ( @xmath2375% at 4.2 k ) to the b - component , a result that does not reflect the 2mm - symmetry of the [ 100 ] direction . considering these and the above observations together with entropy arguments ,
one concludes that the low - temperature structure is either a _ squared - up _ , _ equal - amplitude _ , and collinear state or - as supported by mossbauer analysis- a strongly bunched , _ squared - up _ , and _ equal - amplitude _ state .
as far as the magnon specific heat is concerned , the interactions in both structures can be represented by the hamiltonian of eq.[h ] ( allowing for a unimportant difference in the direction of the axis of quantization ) .
figs.1 - 6 and tab.i demonstrated convincingly that , based on only @xmath4 and @xmath6 the diverse functional form of the measured @xmath73 can be systematized : when both @xmath4 and @xmath5 are large , @xmath152 reflects a magnon contribution from an anisotropic magnetic structure as in @xmath18er , ho , dy , tb . when @xmath4 is relatively small but @xmath5 is large , @xmath108 reflects a magnetic contribution from a quasi - isotropic magnetic structure as in gdni@xmath1b@xmath1c . for a weak @xmath4 and @xmath41 ,
@xmath73 reflects a contribution from a quasi-2@xmath63 structure as in tmni@xmath1b@xmath1c .
the evolution of @xmath4 and @xmath5 across the studied compounds is reasonable .
@xmath5 , on the one hand , reflects predominately the evolution of the de gennes factor ( see eqs.5,7 and tab.i ) as can be appreciated on observing that @xmath5 scales very well with the de gennes factors for @xmath0=tm , er , ho , dy , tb . that the experimentally determined @xmath5 of gdni@xmath1b@xmath1c is a factor of three lower than the one expected from de gennes scaling may be attributed to the additional dependence of @xmath5 on @xmath153 which for gdni@xmath1b@xmath1c is the lowest .
@xmath4 , on the other hand , reflects the combined influence of the anisotropic forces and interlayer coupling .
this is expressed by eq.[gap ] which for the case of , say , erni@xmath1b@xmath1c ( considering @xmath154 @xmath15515 koe and @xmath1560.1 k ) gives a value of 4 k which is close to the experimental value .
the observation that @xmath4 is nonzero for each of the studied compounds is in agreement with the reported anisotropic features of the magnetic and transport properties.@xcite the strong anisotropy of each of @xmath18er , ho , dy , tb is in accord with what is expected from their cef properties .
in contrast , the weak anisotropy observed in gdni@xmath1b@xmath1c is most probably due to a combination of anisotropic exchange , dipolar , and two - ion magnetoelastic couplings .
it is interesting to discuss one particular aspect of the interaction between magnons and superconductivity in , say , @xmath0= ho ( @xmath157 ) and dy ( @xmath17 ) : the magnon - mediated pairbreaking effect that manifests itself in the thermal evolution of @xmath158 [ ref.@xcite ] .
noteworthy , the thermal evolution of @xmath158 of both honi@xmath1b@xmath1c and dyni@xmath1b@xmath1c [ ref.@xcite ] are very similar which , considering the above - mentioned similarity in their magnetic properties , suggests that the involved pairbreaking effects ( in particular the magnon - mediated one ) are similar .
this , in turn , suggests that the magnon characteristic ( say low - energy magnon spectra ) in both compounds must be similar .
this is indeed the case : the analysis of sec.iv.c-d showed that the energy cost for magnon excitation in both compounds is practically equal ( @xmath159 8 k see tab.i ) .
therefore doping of ho into dyni@xmath1b@xmath1c ( up to 80% but still @xmath17 ) would not lead to a softening of @xmath4 . then , for this concentration range , there should be no variation in @xmath26 and @xmath119 even though the degennes factor does vary ( remember that within the antiferromagnetic state , the abrikosov - gorkov pairbreaking mechanism is not valid ) .
this provides an additional experimental confirmation of the hypothesis of cho et al @xcite that the magnon spectrum of ( dy@xmath15ho@xmath14)ni@xmath1b@xmath1c is hardly modified for @xmath1600.8 .
in contrast , for ( dy@xmath15ho@xmath14)ni@xmath1b@xmath1c ( @xmath1610.8 ) , the onset of superconductivity occurs within the paramagnetic state and consequently the dy dopant depresses @xmath119 linearly as expected from the abrikosov - gorkov theory . in summary
, we were able to reveal the magnon specific heat contribution of the heavy members of the borocarbides and to identify the expressions that describe their thermal evolution .
these expressions ( given in terms of only two physically accepted parameters ) were derived from the spin - wave analysis of a simple hamiltonian that consists of effective exchange couplings and anisotropic interactions .
we investigated as well an influence of the magnons on the superconductivity of these af superconductors .
improvements and extension of this analysis are underway .
these include , on the experimental side , probing the magnon contribution in single crystals of @xmath0ni@xmath1b@xmath1c by other ( microscopic and macroscopic ) techniques and , on the theoretical side , a better and more realistic approximation of @xmath45 , cef effects , and magnetoelastic , and anisotropic exchange forces .
p. c. canfield , s. l. budko , b. k. cho , a. lacerda , d. farrell , e. johnston - halperin , v. a. kalatsky and v. l. pokrovsky , phys .
b * 55 * , 970 ( 1997 ) ; a. j. campbell , d. mck.paul and g. j. mcintyre , phys . rev .
b * 61 * , 5872 ( 2000 ) .
a. c. anderson , b. holmstrom , m. kruis and g. r. pickett , phys
. rev . * 183 * , 546 ( 1969 ) ; b. holmstrom , a. c. anderson and m. kruis , phys .
rev . * 188 * , 888 ( 1969 ) ; m. kruis , g. r. pickett and m. c. veuro , solid state commun .
* 14 * , 191 ( 1974 ) . c. song , j. c. lang , c. detlefs , a. letoublon , w. good , j. kim , d.wermeille , s. l. budko , p. c. canfield , a. i. goldman , phys .
b * 64 * , 20403 ( 2001 ) ; c. song , d. wermeille , a. i. goldman , p. c. canfield , j. y. rhee and b. n. harmon , phys . rev .
b * 63 * , 104507 ( 2001 ) ; c. song , z. islam , l. lottermoser , a. i. goldman , p. c. canfield and c. detlefs , phys . rev .
b * 60 * , 6223 ( 2001 ) .
k. levin , m. j. nass , c. ro and g. s. grest , in _ superconductivity in magnetic and exotic materials _
matsubara and a. kotani , ( springer verlag , berlin , 1963 ) , p.104 ; o. fisher , in _ magnetic superconductors _ , edited by k.
h. j. buschow and e. p. wohlfarth , ( elsevier sciences publishers b.v . , 1990 ) , chap.6 , p.465 .
tab.i some zero - field parameters of selected @xmath0ni@xmath1b@xmath1c compounds .
superconducting @xmath119 , magnetic @xmath99 , _ magnetic structure _ , propagation _ wave vector _ , and _ moment direction _ are taken from ref .
@xcite . squaring of the modulated sdw state is taken to occur at lower temperatures .
the gap , @xmath8 and characteristic temperature , @xmath6 were determined from the indicated equation and figure .
@xmath162 of tmni@xmath1b@xmath1c is calculated by substituting into eqs.[c - definition],[theta ] the fit values ( @xmath42=.8 k and @xmath163=.2 ) given by movshvich et al.@xcite .
@xmath164 is the de gennes scaling of @xmath5 taking that of @xmath18ho as a reference .
[ c]|l|l|l|l|l|l|l|l|l|l|l|l|@xmath0 & deg & @xmath119 & @xmath99 & magnetic & wave & moment & @xmath4 & @xmath162 & @xmath164 & eq .
# + & & k & k & structure & vector & direction & @xmath1040.2 k & @xmath1040.2 k & k & & + gd & 15.75 & 0 & 19.5 & sdw & [ .55,0,0 ] & [ 0,1,0 ] & 1.9 & 12.5 & 34.0 & [ cm - bessel ] & 1 + tb & 10.5 & 0 & 15.4 & sdw & [ .555,0,0 ] & [ 1,0,0 ] & 7 & 21.5 & 22.6 & [ cm - bessel ] & 2 + dy & 7.08 & 6 & 9.5 & 3d , af & [ 0,0,1 ] & [ 1,1,0 ] & 8.3 & 19.3 & 15.3 & [ cm - bessel ] & 3 + ho & 4.5 & 8 & 5 & 3d , af & [ 0,0,1 ] & [ 1,1,0 ] & 8.3 & 9.7 & 9.7 & [ cm - bessel ] & 5 + er & 2.5 & 10.5 & 5.9 & sdw & [ .553,0,0 ] & [ 0,1,0 ] & 5.4 & 7.4 & 5.5 & [ cm - bessel ] & 6 + tm & 1.16 & 11 & 1.5 & sdw & [ .093,.093,0 ] & [ 0,0,1 ] & 0 & 3.8 & 2.5 & [ cm-2d ] & 4 in ref.@xcite + tab.ii list of values of @xmath89 , @xmath165 and @xmath5 of @xmath0ni@xmath1b@xmath3c ( @xmath0=gd , tb , er , ho , dy , tm ) . also given are the expressions used for evaluating @xmath166 the nuclear hyperfine parameters @xmath167 and @xmath168 ( see refs.@xcite ) are also indicated : the upper ( lower ) values of dyni@xmath1b@xmath1c correspond to the isotope @xmath169dy ( @xmath170dy ) .
data of tmni@xmath1b@xmath1c were taken from ref.@xcite .
[ c]|l|l|l|l|l|l|l|@xmath0 & @xmath89 & @xmath171 & @xmath5 ( @xmath172 & & fig . # + & mj / molek@xmath86 & j / molek & k ( mj / molek@xmath88 ) & @xmath167 ( k ) & @xmath168(k ) & + gd & 17.5 & @xmath173 & 392(0.1935 ) & 0 & 0 & 1 + tb & 17.5 & @xmath173 & 391(0.196 ) & 0.14(2 ) & 0.02(1 ) & 2 + dy & 17.5 & 3@xmath174 & 388 ( 0.200 ) & @xmath175{l}% $ ] -.0396@xmath176 & @xmath175{l}% $ ] .009@xmath177 & 3 + ho & 17.5 & 3@xmath174 & 386 ( 0.203 ) & @xmath178 & @xmath179 & 5 + er & 17.5 & 3@xmath174 & 384 ( 0.206 ) & @xmath180 & @xmath181 & 6 + tm & @xmath2318 & @xmath173 & @xmath23355(@xmath230.26 ) & @xmath182 & @xmath183 & 4 in ref.@xcite +
|
zero - field specific heat of the single crystals @xmath0ni@xmath1b@xmath1c
( @xmath0= er , ho , dy , tb , gd ) was measured within the temperature range 0.1 k@xmath225 k. linearized spin wave analysis was successfully applied to account for and to rationalize the thermal evolution of the low temperature magnetic specific heats of all the studied compounds ( as well as the one reported for tmni@xmath3b@xmath1c ) in terms of only two parameters , namely an energy gap @xmath4 and a characteristic temperature @xmath5 .
the evolution of @xmath5 and @xmath4 across the studied compounds correlates very well with the known magnetic properties . @xmath6 as a measure of the effective @xmath7 exchange couplings , scales reasonably well with the de gennes factor .
@xmath8 on the other hand , reflects predominately the anisotropic properties : @xmath9 k for gdni@xmath1b@xmath1c , @xmath10 k for erni@xmath1b@xmath1c , @xmath11 k for tbni@xmath3b@xmath1c , and @xmath12 k for each of honi@xmath1b@xmath1c and dyni@xmath1b@xmath3c .
the equality in @xmath4 of honi@xmath1b@xmath1c and dyni@xmath1b@xmath1c , coupled with the similarity in their magnetic configurations , indicates that a variation of @xmath13 in the solid solution ho@xmath14dy@xmath15ni@xmath1b@xmath1c ( @xmath16 and @xmath17 ) would not lead to any softening of @xmath4 .
this supports the hypothesis of cho et al ( prl * 77*,163(1996 ) ) concerning the influence of the collective magnetic excitations on the superconducting state .
this work underlines the importance of spin - wave excitations for a valid description of low - temperature thermodynamics of borocarbides .
|
since the original paper by ising @xcite , magnetic models on a lattice have been extensively used over the years to investigate the statistical physics of interacting many - body systems . in particular , many generalizations of the ising model have been proposed . among them , also models with long - range interactions .
the thermodynamics and the dynamics of systems of particles interacting with long - range forces are particularly interesting because display a series of anomalies with respect to systems with short - range interactions@xcite . by long - range interaction
it is usually intended that the modulus of the potential energy decays , at large distance , not faster than the inverse of the distance to the power of the spatial dimension .
the main reason of the observed anomalies is that systems with long - range forces in general violate _
extensivity _ and _ additivity _ , two basic properties to derive the thermodynamics of a system .
extensivity means that the thermodynamic potentials scale with the system size , i.e. that the specific thermodynamic potentials ( the thermodynamic potentials per particle ) do not diverge in the thermodynamic limit .
additivity means that if we divide the system into two macroscopic parts , the thermodynamic potentials of the whole system are , in the thermodynamic limit , the sum of those of the two components .
though extensivity can be artificially restored by an ad - hoc introduction of a n - dependent coupling constant ( the so called kac s prescription @xcite ) in the interaction , the problem of non - additivity is still present @xcite .
thus , the study of long - range magnetic systems on lattices can give insights on the statistical properties of long - range potentials , and is therefore useful to investigate the statistical mechanics and the dynamics of such systems .
in this paper we study the @xmath0 model , a model of planar spins on a d - dimensional lattice with couplings that decay as the inverse of the distances between spins raised to the power @xmath1 @xcite .
when @xmath1 is smaller than @xmath2 additivity does not hold and the system shows a series of anomalies .
even ensemble equivalence , whose proof is based on the possibility of separating the energy of a subsystem from that of the whole , might not be guaranteed in that limit .
in particular we focus on @xmath3 .
in such a case the model reduces to a mean field model@xcite , since a spin interacts equally with all the others independently of their position on the lattice .
this case is extremely important since it has been proved that all the cases with @xmath4 @xcite can be reduced to it .
moreover the dynamical behavior of the model with @xmath3 is also a representative example of other nonextensive systems @xcite .
we will discuss the equilibrium phase and the dynamical anomalies found in an energy region before the second - order critical point when one studies the relaxation towards equilibrium .
the connections with tsallis generalized thermodynamics @xcite and with glassy systems@xcite will be also addressed .
the paper is organized as follows .
we introduce the model in section 2 , the equilibrium thermodynamics is presented in section 3 , while the anomalous dynamical behavior and its possible theoretical interpretation is discussed in section 4 .
conclusions and future prespectives are presented in section 5 .
the @xmath5 model has been introduced in @xcite and describes a system of classical bidimensional spins ( xy spins ) with mass @xmath6 .
the hamiltonian is : @xmath7 with @xmath8 .
the @xmath9 spins are placed at the sites of a generic @xmath2-dimensional lattice , and each one is represented by the conjugate canonical pair @xmath10 , where the @xmath11 s are the momenta and the @xmath12 s @xmath13 are the angles of rotation on a family of parallel planes , each one defined at each lattice point .
the interaction between rotators @xmath14 and @xmath15 decays as the inverse of their distance @xmath16 to the power @xmath17 .
such hamiltonian is extensive if the thermodynamic limit ( tl ) @xmath18 of the canonical partition function @xmath19 exists and is finite .
this is assured for each @xmath1 by the presence of the rescaling factor @xmath20 in front of the double sum of the potential energy .
@xmath20 is a function of the lattice parameters @xmath21 which is proportional to the range @xmath22 of the interaction defined by @xcite : @xmath23 the sum is independent of the origin @xmath14 because of periodic conditions .
then , for each @xmath1 , @xmath24 is proportional to @xmath9 .
when @xmath25 , which we call here the _ short - range _ case , @xmath22 is finite in the tl @xcite , and things go as if each rotator interacted with a finite number of rotators , those within range @xmath22 . on the contrary when @xmath26 , which we consequently call the _ long - range _ case , @xmath22 diverges in the tl and the factor @xmath27 in ( [ hmfalpha ] ) compensates for this .
+ in fig.1 we report temperature and magnetization vs the energy density .
the full curves are the exact results obtained in the canonical ensemble , the boltzmann - gibbs ( bg ) equilibrium .
the numerical simulations ( symbols ) were performed at fixed total energy ( _ microcanonical _ molecular dynamics ) by starting the system close to the equilibrium , i.e. with an initial gaussian distribution of angles and velocities , and reproduce the theoretical curves .
this is already true for small system sizes as @xmath78 , apart from some finite size effects in the homogeneous phase @xcite .
as we shall see in the next section the scenario is very different when the system is started with strong out - of - equilibrium initial conditions : in such a case the dynamics does have difficulties in reaching the bg equilibrium and shows a series of anomalies .
the dynamics of each particle obeys the following pendulum equation of motion : @xmath79 where @xmath34 and @xmath35 have a non trivial time dependence , related to the motion of all the other particles in the system .
equations ( [ pend ] ) can be integrated numerically , see for example refs . @xcite for technical details . in this section
we show that , in an energy range from @xmath80 up to @xmath81 , when the system is started with strong out - of - equilibrium initial conditions , the model has a non trivial dynamical relaxation to the boltzmann - gibbs ( bg ) equilibrium .
the class of out - of - equilibrium initial conditions we consider , called _ water bag _ initial conditions , consists in @xmath82 and the momenta uniformly distributed ( according to the total energy density @xmath83 ) . in fig.2 we report , for u=0.69 and n=500 , the time evolution of @xmath84 ( where @xmath85 denotes the time averaged kinetic energy ) , a quantity that coincides with the temperature t. as expected , the system does not relax immediately to the bg equilibrium , but rapidly reaches a quasi - stationary state ( qss ) corresponding to a temperature plateau situated below the canonical prediction ( dotted curve ) .
the t vs u plot ( caloric curve ) , shown in inset ( c ) for the qss regime , confirms a large disagreement with the equilibrium prediction ( inset d ) .
furthermore , it turns out that the system remains trapped in such a state for a time that diverges with the size @xmath9 of the system @xcite .
this means that , if the thermodynamic limit is performed before the infinite - time limit , the qss become stable and the system never relax to the bg equilibrium , exhibiting different equilibrium properties characterized by non - gaussian velocity distributions ( see inset a ) @xcite .
such velocity distributions have been fitted in ref.@xcite and have been shown to be in agreement with the prediction of the tsallis generalized thermodynamics @xcite .
the latter is a theoretical formalism well suited to describe all those situations where long - range correlations , weak mixing and fractal structures in phase space are present @xcite , such as for example excited plasmas @xcite , turbulent fluids @xcite , maps at the edge of chaos @xcite , high energy nuclear reactions @xcite and cosmic rays fluxes@xcite .
+ the characteristics of the qss have been studied in several papers : vanishing lyapunov spectrum @xcite , negative microcanonical specific heat@xcite , dynamical correlations in phase - space @xcite , lvy walks and anomalous diffusion@xcite .
recently also the validity of the zeroth principle of thermodynamics has been numerically demonstrated for these metastable states@xcite .
+ in the following of this paper we focus on the study of anomalous diffusion , long - range correlations and the spin - glass phase .
the link between relaxation to the bg equilibrium , and anomalous diffusion was studied in ref.@xcite .
notice that diffusion starts around @xmath102 when particles start to be evaporated from the main cluster @xcite .
this result can be interpreted in the following way . when started with water - bag initial condition ( @xmath103 )
the system immediately decays into the qss plateau with @xmath104 . in fig.2
this initial part is not shown .
then a very slow microscopic relaxation occurs since the force exerted on each single spin / particle is almost zero @xcite . during the slow relaxation process ,
many small rotating clusters on the unitary circle are continuosly formed and compete each other by trapping the particles in order to reach a configuration compatible with the final bg equilibrium state .
thus , particles remains trapped for a while in the clusters , until they finally succeed to escape again and so on . trapping times and escape times obey power - law decays with characteristic exponents that allow to relate anomalous diffusion with lvy walks .
@xcite for more details .
such a mechanism of trappings and lvy walks disappears when the system finally reaches equilibrium since at that time only one large cluster is present ( the system is below the critical point in the ferromagnetic phase ) .
anomalous diffusion can be obtained within a generalized fokker - planck equation which generates tsallis distributions with an entropic index @xmath105 , see refs.@xcite .
one can extract the following relationship between the exponent @xmath106 which characterizes anomalous diffusion and the entropic index @xmath105 of tsallis thermostatistics @xmath107 considering the value of @xmath108 observed in our case , we would expect then a value of the entropic index @xmath109 which should characterize the dynamical anomalies of the hmf model .
actually , this is not the value found in ref.@xcite for the velocity pdfs , where we obtained only an effective entropic index .
however this value is in good agreement to what has been found for the decay of the velocity correlation functions which will be examined in the next section .
the global effects of the competition between magnetic clusters discussed in the previous subsection prevents the system from exploring all the available phase space .
thus one observes a sort of _ dynamical frustration _ which suggests also an interesting connection with the _ weak - ergodicity breaking _
scenario typical of glassy systems .
such a scenario , introduced by bouchaud et al .
@xcite , generally occurs when the phase - space of the system we consider is not _ a - priori _ broken into mutually inaccessible regions , but the system remains confined only into a restricted part of it : consequentely one finds slowly decaying correlation functions and aging , i.e. the presence of strong memory effects that depend on the history of the system .
this is just what happens in the hmf model .
+ the velocity autocorrelation functions in the qss regime have been studied in ref@xcite . in its simplest form
the autocorrelation function of the particles velocity can be written as @xcite @xmath110 if we want to take into account several dynamical realizations ( events ) in order to obtain better averaged quantities , it is possible to use the following alternative definition of the autocorrelation function @xmath111 where @xmath112 is the n - component velocity vector and the brackets @xmath113 indicate the average over different events , while @xmath114 and @xmath115 are the standard deviations at time @xmath38 and at the initial time .
such a function has been numerically evaluated @xcite in the qss region .
the results are shown in fig.4 for the case @xmath100 and @xmath116 : a power - law tail is observed , this being a signature of long - range correlations . to study the aging phenomenon we need to evaluate a two points correlation function : @xmath124 where @xmath125 is the waiting time .
this function has been evaluated in ref.@xcite . in the qss region ,
one finds a strong dependence on the waiting time @xmath125 see fig.5 ( a ) for the case @xmath100 and @xmath116 .
the curves obey a precise scaling as that one observed for glasses . rescaling the time of the various correlation curves by @xmath126 with @xmath127
, we get a unique curve with a very interesting power - law tail , see fig.5 ( b ) . also in this case as in the previous figure
, one can reproduce nicely the time evolution of the rescaled correlation function with a q - exponential curve .
the best fit , reported in the figure , was obtained for case @xmath128 , @xmath129 and @xmath130 .
the q - logarithm is reported in panel ( c ) in order to compare the quality of the fit .
notice that this value of @xmath105 is not very different from the one extracted from the decay of the autocorrelation function ( 22 ) and from that one extracted from the anomalous diffusion . a more detailed investigation in order to deduce analitically
the predicted value of the entropic index from the observed anomalies is in progress . in the long - range spin - glass models the aging phenomenon below a transition temperature
is associated with the complex energy landscape characteristic of the frustrated models .
the latter is rich of metastable states that play the role of dynamical traps which can confine the system for a long time and cause the observed history - dependent slow relaxation dynamics@xcite .
the parallel with spin - glass systems can be made stronger by the introduction of a new order parameter for the qss plateau , the @xmath131 @xcite .
the polarization represents the temporal average ( over a time interval @xmath132 inside the qss plateau ) of the spin vectors , whose modulus is also averaged over the @xmath9 spins , i.e. @xmath133 such an order parameter allows to characterize in a quantitative way the dynamical freezing and frustration of the qss regime by interpreting the latter as a spin - glass phase .
in fact , in analogy with the spin - glass phase of the long - range sherrington - kirkpatrick ( sk ) model @xcite , we have observed that , in the qss regime and in the thermodynamic limit , the magnetization vanishes as @xmath134 , but the polarization @xmath135 remains constant around a value equal to @xmath136 , see fig.2 of ref .
therefore @xmath135 allows to distinguish between the disorderd qss glassy phase and the high temperature one .
this result is very interesting because , at variance with standard glassy systems , neither disorder nor frustration are present _ a - priori _ in the elementary interactions of the hmf model . on the contrary , they emerge naturally as dynamical features of the model in qss region and , like the other features seen before , disappear when the system reaches finally the bg equilibrium .
the introduction of the polarization order parameter , which connects the qss dynamical frustration with the edwards - anderson order parameter of the sk model , opens a new perspective to understand the true nature of metastability in long - range hamiltonian systems in terms of an emerging glassy behavior .
in this paper we have briefly reviewed the dynamics and thermodynamics of the hamiltonian mean field model .
this is a model that allows to study long - range interactions in many - body hamiltonian systems .
the equilibrium properties of the hmf model can be derived analitically , though the model dynamics presents very interesting anomalies like metastability , superdiffusion , non - gaussian velocity pdfs , long - range correlations , vanishing lyapunov exponents and weak - ergodicity breaking in an energy region below the critical point .
these anomalies are common to many long - range systems .
recently it has been found that such anomalous regime can be characterized as a glassy phase .
therefore the model represents also a very interesting bridge between nonextensive systems and glassy systems and in this respect it will deserve more detailed investigations in the future .
the link with the nonextensive thermostatistics formalism proposed by constantino tsallis is also a very promising and intriguing line of research .
all the anomalies point in that direction and the future years will be crucial in order to establish this connection in a firm and rigorous way as it has recently been done for example in unimodal maps .
m. antoni and s.ruffo , phys .
e * 52 * ( 1995 ) 2361 ; v. latora , a. rapisarda and s. ruffo , phys .
* 80 * ( 1998 ) 692 ; physica d * 131 * ( 1999 ) 38 and progr .
* 139 * ( 2000 ) 204 .
see also t. dauxois , v. latora , a. rapisarda , s. ruffo and a. torcini , in ref.@xcite , p.458 . c. tsallis , j. stat
* 52 * ( 1988 ) 479 ; for recent reviews see : c. tsallis , _ nonextensive statistical mechanics and thermodynamics _ , lecture notes in physics , eds .
s. abe and y. okamoto , springer , berlin , ( 2001 ) ; proceedings of next2001 , special issue of physica a 305 ( 2002 ) eds .
g. kaniadakis , m. lissia and a rapisarda ; _ nonextensive entropy- interdisciplinary applications _ , c. tsallis and m. gell - mann eds . , oxford university press ( 2004 ) .
see also http://tsallis.cat.cbpf.br/biblio.htm for a regularly updated and complete bibliography .
for the recent debate about nonextensive thermodynamics see also : a. cho , science * 297 * ( 2002 ) 1268 ; s. abe and a.k .
rajagopal , science * 300 * ( 2003 ) 249 ; a. plastino , science * 300 * ( 2003 ) 250 ; v. latora , a. rapisarda and a. robledo , science * 300 * ( 2003 ) 250 . c. beck , g.s .
lewis and h.l .
swinney , phys .
e * 63 * ( 2001 ) 035303(r ) .
see also c. beck , physica a * 295 * ( 2001 ) 195 , phys .
a * 287 * ( 2001 ) 240 and phys .
* 87 * ( 2001 ) 180201 and physica a * 305 * ( 2002 ) 209 ; t. arimitsu and n. arimitsu , physica a * 305 * ( 2002 ) 218 .
lyra and c. tsallis , phys .
* 80 * ( 1998 ) 53 ; u. tirnakli , c. tsallis and m.l .
lyra , eur .
j. b * 10 * ( 1999 ) 309 ; f.a.b.f .
de moura , u. tirnakli and m.l .
lyra phys .
e * 62 * ( 2000 ) 6361 .
latora , baranger , rapisarda and tsallis , phys .
lett . a * 273 * ( 2000 ) 97 ; u. tirnakli , phys .
e * 62 * ( 2000 ) 7857 and phys .
e * 66 * ( 2002 ) 066212 ; u. tirnakli u , c. tsallis , m. l. lyra , phys .
e * 65 * ( 2002 ) 036207 ; e.p .
borges , c. tsallis , g.f.j .
ananos , p.m.c .
de oliveira , phys .
* 25 * ( 2002 ) 254103 .
i. bediaga , e.m.f .
curado , j. miranda , physica a * 286 * ( 2000 ) 156 ; g. wilk and z. wlodarczyk , phys .
* 84 * ( 2000 ) 2770 ; o.v .
utyuzh , g. wilk g and z. wlodarczyk j. of phys .
g * 26 * ( 2000 ) l39 ; w.m .
alberico , a .
lavagno and p. quarati , eur .
j. c * 12 * ( 1999 ) 499 .
|
the dynamics and the thermodynamics of particles / spins interacting via long - range forces display several unusual features with respect to systems with short - range interactions . the hamiltonian mean field ( hmf ) model , a hamiltonian system of n classical inertial spins with infinite - range interactions represents a paradigmatic example of this class of systems .
the equilibrium properties of the model can be derived analytically in the canonical ensemble : in particular the model shows a second order phase transition from a ferromagnetic to a paramagnetic phase .
strong anomalies are observed in the process of relaxation towards equilibrium for a particular class of out - of - equilibrium initial conditions .
in fact the numerical simulations show the presence of quasi - stationary state ( qss ) , i.e. metastable states which become stable if the thermodynamic limit is taken before the infinite time limit .
the qss differ strongly from boltzmann - gibbs equilibrium states : they exhibit negative specific heat , vanishing lyapunov exponents and weak mixing , non - gaussian velocity distributions and anomalous diffusion , slowly - decaying correlations and aging .
such a scenario provides strong hints for the possible application of tsallis generalized thermostatistics .
the qss have been recently interpreted as a spin - glass phase of the model .
this link indicates another promising line of research , which is not alternative to the previous one .
leer.eps gsave 72 31 moveto 72 342 lineto 601 342 lineto 601 31 lineto 72 31 lineto showpage grestore
|
radio frequency ( rf ) signal enabled wireless energy transfer ( wet ) has become a promising technology to provide perpetual , cost - effective , and convenient energy supply to low - power wireless devices such as rf identification ( rfid ) tags and sensor nodes for various low - power applications in future smart commercial and industrial systems with wireless internet - of - things ( iot ) @xcite . in rf - based wet systems , dedicated energy transmitters ( ets )
are deployed to control and coordinate the transfer of wireless energy via rf signals to a set of distributed energy receivers ( ers ) , and each er uses the rectifier at each of its receive antennas ( also known as rectennas as a whole ) to convert the received rf signals to direct current ( dc ) signals , which are combined to charge the battery at the er @xcite ( see , e.g. , a point - to - point wet system in fig .
[ fig : system ] ) .
examples of commercial ers include the p2110 and p2210 powerharvester receivers developed by the powercast company @xcite .
due to the long operating range ( say , tens of meters ) and flexibility to charge multiple devices at the same time ( thanks to the broadcast property of radio signal ) , rf - based far - field wet has competitive advantages over existing near - field wet techniques such as induction coupling , which generally operates with much shorter distance ( say , centimeters ) and for point - to - point power transfer only @xcite .
furthermore , since rf signal can carry energy as well as modulated information at the same time , a joint investigation of wireless information and energy transfer is appealing , which has recently drawn a great amount of attention .
in particular , simultaneous wireless information and power transfer ( swipt ) ( see , e.g. , @xcite ) and wireless powered communication network ( wpcn ) ( see , e.g. , @xcite ) are two important operation models that have been thoroughly studied in literature . in swipt systems , wireless information and energy
are transmitted using the same signal ; while in wpcn , wireless information transmission is powered by wet using different signals .
= 1 one major practical challenge for implementing rf - based wet is the significant decay of energy transfer efficiency of wet over the distance from the et to the er due to the severe prorogation power loss of rf signal . to improve the efficiency , multiple - input multiple - output ( mimo ) techniques by equipping multiple antennas at the et and/or the
er have been proposed as an appealing solution ( see , e.g. , @xcite and the references therein ) .
multiple antennas at the et help focus the transmitted wireless energy on the direction to the er via the technique of digital beamforming or so - called energy beamforming , while multiple antennas at the er increase the effective aperture area for received rf energy harvesting , both leading to significantly enhanced energy transfer efficiency .
this helps extend the operating range of wet with given ers energy requirements , and/or enable more power - consuming applications by increasing ers harvested energy with fixed distances . however , the benefit of energy beamforming in mimo wet critically relies on the availability of the channel state information ( csi ) at the et , which needs to be efficiently learned in practical systems .
there are in general three types of channel acquisition methods that have been proposed in the literature for mimo wet .
the first method is by exploiting the channel reciprocity between the forward ( from the et to the er ) and reverse ( from the er to the et ) links @xcite , where the et obtains the forward link csi by performing a reverse link channel estimation based on the training signals sent by the er . it is worth noting that this method is only applicable for time - division duplex ( tdd ) based systems , where the condition of channel reciprocity practically holds with good accuracy .
the second method is via sending training signals directly in the forward link from the et to the er , through which the er estimates the mimo channel and then sends the estimated channel back to the et via the reverse link @xcite .
this method is reminiscent of the conventional channel estimation and feedback approach in wireless communication ( see , e.g. , @xcite and the references therein ) , which is applicable for both tdd and frequency - division duplex ( fdd ) based systems .
however , under the new setup of wet ( instead of wireless communication ) , this method requires the er ( not an information receiver ) to implement additional baseband signal processing for channel estimation , which is not feasible with the existing hardware at the er designed only for rf energy harvesting ( instead of information decoding ) as shown in fig .
[ fig : system ] . to cater for the hardware limitation of existing ers , in our previous work @xcite
we proposed a third channel learning method for mimo wet based on the energy measurement feedback by the er . specifically , the er measures its harvested energy levels over different training intervals using an energy meter ( see fig .
[ fig : system ] ) , and sends one feedback bit to the et per interval to indicate the increase or decrease of the measured energy in the current interval as compared to that in the previous interval .
based on the feedback bits collected in the present and past intervals , the et adjusts its transmit beamforming in subsequent training intervals and obtains refined estimates of the mimo channel .
note that similar channel learning methods with one - bit feedback have also been studied in cognitive radio mimo systems @xcite .
this energy feedback based channel learning method is applicable for both tdd and fdd systems , and can be implemented without additional baseband processing modules at the er . however , the existing studies in @xcite considered only one feedback bit per training interval ; while it remains unknown how to efficiently implement this method for the general setup with multiple feedback bits available per interval .
this thus motivates our study in this work .
in this paper , we propose a general channel learning design framework for mimo wet based on the er s energy feedback over a finite - rate reverse link from the er to the et ( thus including the one - bit feedback per interval designed in @xcite as a special case ) .
note that our results can also be similarly applied to other applications for channel learning in wireless communications with received signal power / energy based feedback ( such as cognitive radio systems in @xcite ) .
we consider a quasi - static channel model and a block - based energy transmission with each block spanning over a constant mimo channel for wet .
each block is further divided into two phases for channel learning and energy transmission , respectively . in the channel learning phase
, the et learns the mimo channel progressively by adjusting its training beamforming weights according to the received energy feedback from the er .
it is assumed that at each feedback interval the er sends @xmath0 bits to the et via the reverse link . in general , the feedback bits at each interval are generated by the er via efficiently encoding the measured energy levels at the present and past intervals subject to the given feedback rate constraint . based on the estimated mimo channel from the channel learning phase , in the subsequent energy transmission phase , the et implements the optimal transmit energy beamforming to maximize the energy transferred to the er . in this paper , we focus our study on the channel learning phase by investigating the energy feedback design at the er and the training beamforming design and channel estimation method at the et .
our main results are summarized as follows . *
first , we present a general design framework for the energy measurement feedback based channel learning with limited feedback rate .
similar to @xcite for the special case of one - bit feedback ( i.e. , @xmath1 ) , our design is based on the technique of analytic center cutting plane method ( accpm ) in convex optimization @xcite .
the basic idea of accpm is as follows : based on the @xmath2 feedback bits from the er at each interval , the et constructs one or more linear inequalities on the mimo channel matrix to be estimated , where each inequality is used as a cutting plane for helping localize the mimo channel in a multi - dimensional complex matrix space .
based on these cutting planes , the et also designs the transmit beamforming for subsequent training intervals and obtains refined estimates of the mimo channel .
* under this general framework , we then propose two specific feedback designs , namely , energy quantization and energy comparison , where the @xmath2 feedback bits per interval are generated at the er by quantizing the measured energy level at the current interval and comparing it with those in the @xmath2 previous intervals , respectively .
note that when @xmath1 , the energy comparison based channel learning scheme degenerates to that with one - bit feedback per interval as in our previous work @xcite .
* finally , we provide extensive numerical results to compare the performance of our proposed feedback and channel learning schemes in terms of both mean - squared error of the estimated mimo channel norm as well as the resulting energy beamforming gain .
it is shown that energy quantization generally preforms better than energy comparison when @xmath2 is large , while energy comparison is more effective vice versa .
the remainder of this paper is organized as follows .
section [ sec : sysmod ] introduces the system model and the two - phase transmission protocol .
section [ sec : framework ] presents the general design framework for the energy feedback based channel learning using accpm .
sections [ sec : quantization : feedback ] and [ sec : comparison : feedback ] present the energy quantization and energy comparison based feedback and channel learning schemes , respectively .
section [ sec : low : complexity ] presents a complexity reduction approach for the proposed schemes by pruning irrelevant cutting planes in accpm .
section [ sec : numerical : results ] provides numerical results to evaluate the performance of proposed schemes .
finally , section [ sec : conclusion ] concludes the paper .
_ notation : _ boldface letters refer to vectors ( lower case ) or matrices ( upper case ) . for a square matrix @xmath3 , @xmath4 and
@xmath5 denote its determinant and trace , respectively , while @xmath6 means that @xmath3 is positive semi - definite . for an arbitrary - size matrix @xmath7 ,
@xmath8 , @xmath9 , @xmath10 , and @xmath11 denote the frobenius norm , rank , conjugate transpose and transpose of @xmath7 , respectively . @xmath12 and @xmath13 denote an identity matrix and an all - zero matrix , respectively , with appropriate dimensions . @xmath14 and @xmath15 denote the space of @xmath16 complex and real matrices , respectively .
@xmath17 denotes the statistical expectation .
@xmath18 denotes the euclidean norm of a complex vector @xmath19 , and @xmath20 denotes the magnitude of a complex number @xmath21 .
@xmath22 denotes the largest integer not greater than the real number @xmath23 , and @xmath24 denotes the smallest integer not less than @xmath23 , respectively .
symbol @xmath25 denotes the complex number @xmath26 .
we consider a point - to - point mimo system for wet as shown in fig .
[ fig : system ] , in which one et with @xmath27 transmit antennas sends wireless energy via rf signal to one er with @xmath28 receive antennas .
we assume a quasi - static flat - fading channel model , in which the wireless channel remains constant over each block of our interest and may change from one block to another .
let the block length be denoted by @xmath29 , which is assumed to be sufficiently long for typical wet applications with low mobility receivers .
we consider transmit beamforming at the et for wet . without loss of generality
, we assume that there are a total of @xmath30 energy beams employed at the et , where @xmath31 is a design parameter that will be specified later .
we denote the transmitted energy signal at the et as @xmath32 , where @xmath33 denotes the @xmath34th transmit beamforming vector and @xmath35 represents the corresponding energy - bearing signal . since the energy signal @xmath35 s do not carry any information , they are assumed to be independent pseudorandom sequences with zero mean and unit variance , i.e. , @xmath36 , in order to spread the signal power over the frequency to satisfy certain power spectral mask constraint imposed by the regulations on microwave radiation .
accordingly , the transmit covariance matrix at the et is given by @xmath37 . note that for any given transmit covariance matrix @xmath3 , the corresponding energy beamforming vectors , i.e. , @xmath38 , can be obtained via performing the eigenvalue decomposition ( evd ) on @xmath3 , where @xmath39 .
in addition , we assume that the maximum transmit sum - power of the et is given by @xmath40 , and thus we have @xmath41 . with energy beamforming at the et , the er can harvest the energy carried by the @xmath30 energy beams from all the @xmath42 receive antennas .
let @xmath43 denote the mimo channel matrix from the et to the er , and define @xmath44 .
due to the law of energy conservation , the harvested rf - band power from all receive antennas at the er is proportional to o that of the equivalent baseband signal @xcite . as a result ,
the harvested energy at the er in one particular block of interest is expressed as @xmath45 where @xmath46 denotes the energy harvesting efficiency at the er . since @xmath47 is a constant , we normalize it as @xmath48 in the sequel of this paper unless otherwise stated . in this subsection
, we present the energy signal design with or without csi at the et .
first , consider that the et perfectly knows the csi . in this case , the et designs the energy beamforming to maximize its transferred energy to the er subject to the given transmit sum - power constraint .
mathematically , the transferred energy maximization problem is formulated as @xmath49 let the @xmath50 eigenvalues of @xmath51 be denoted by @xmath52 , and the corresponding eigenvectors by @xmath53 , respectively .
then it is shown in @xcite that the optimal solution to problem ( [ eqn : problem2 ] ) is given by @xmath54 , which achieves the maximum transferred energy to the er given by @xmath55 here , it follows that @xmath56 ; as a result , only one single transmit energy beam is needed to achieve the optimality .
such a single - beam solution is termed as _ optimal energy beamforming ( oeb)_. on the other hand , consider the case when the et does not have any knowledge about the csi . in this case , the oeb is not implementable .
alternatively , the et can employ an isotropic transmission to radiate its transmit power uniformly over all directions , with the corresponding transmit covariance matrix given by @xmath57 .
correspondingly , the harvested energy at the er is expressed as @xmath58 by comparing @xmath59 in ( [ eqn : qoeb ] ) and @xmath60 in ( [ eqn : qiso ] ) , it is observed that the oeb with perfect csi at the et offers an energy beamforming gain over the isotropic transmission without csi , which can be expressed as @xmath61 due to the fact that @xmath52 , it is evident that @xmath62 , where @xmath63 holds when @xmath64 ( with @xmath65 ) , while @xmath66 is attained when @xmath67 and @xmath68 ( i.e. , @xmath69 ) . for practical wet systems with @xmath70
, we normally have @xmath71 ( and @xmath72 if @xmath73 ) , since the wireless channel is often line - of - sight ( los ) dominant given the limited transmission range of most interest ( say , several meters ) .
in this case , an energy beamforming gain @xmath74 can be achieved with the availability of csi at the et .
= 1 to achieve the promising energy beamforming gain in practice , we consider a two - phase transmission protocol for the et to learn the mimo channel and based on the estimated channel implement the oeb . as shown in fig .
[ fig : protocol ] , each transmission block is divided into two phases , for the purposes of channel learning and energy transmission , respectively .
we explain these two phases in detail as follows .
first , the channel learning phase corresponds to the first @xmath75 amount of time in each block , @xmath76 , which is further divided into @xmath77 training intervals .
let @xmath78 denote the length of each training interval .
then we have @xmath79 . in each of the @xmath80 intervals ,
the et transmits different sets of energy beams ( each set corresponding to a different transmit covariance matrix ) to the er . let the transmit covariance matrix at the et in the interval @xmath81 be denoted by @xmath82 , where @xmath83 and @xmath84 .
then the harvested energy by the er in the @xmath85-th interval is given by @xmath86 specifically , the et implements energy measurement via the energy meter shown in fig .
[ fig : system ] in the first @xmath87 amount of time in each interval , number of multiple - input single - output ( miso ) channels each for one receive antenna .
this can increase the channel learning convergence rate at the cost of higher hardware and feedback costs at the er . ] and then sends back the measured energy information to the et at the remaining @xmath88 amount of time , where @xmath89 . the harvested energy amount measured by the er in the @xmath85-th interval
is denoted by @xmath90 it is assumed that the er can send @xmath2 bits to the et at each interval via the reverse link , where @xmath91 is an integer with @xmath92 denoting the feedback rate from the er to the et .
the value of @xmath2 is practically finite since the er is energy - constrained for sending training signals and the available bandwidth for feedback may be shared among multiple ers ( to their respective ets ) .
we denote the @xmath2-bits feedback at each interval @xmath85 as @xmath93 . in general
, @xmath93 specifies the energy feedback scheme in use at the er which is a mapping from its present and past energy observations ( i.e. , @xmath94 ) to the @xmath2 feedback bits at interval @xmath95 .
furthermore , let @xmath96 and @xmath97 denote the energy consumed by the er for the energy measurement and the information feedback in each interval , respectively .
note that the values of @xmath87 and @xmath88 should be chosen in practice by considering the sensitivity of the energy meter at each er and the feedback link capacity from each er to the et , respectively , as well as the er s energy consumption for energy measurement and feedback . for the purpose of exposition
, we make the following assumptions on the er s energy measurement and feedback . *
the energy measurement of each @xmath98 at the er is perfectly accurate , while the effect of energy measurement error at the er on the channel learning performance will be shown by simulation in section [ sec : energymeasurementerror ] . *
the feedback link is ideal and each feedback @xmath93 is received by the et at the end of interval @xmath85 without any error or delay ; based on the feedback @xmath99 , the et obtains a final estimate of the mimo channel at the end of the channel learning phase .
the details of energy feedback design at the er and training signal design and channel estimation method at the et will be given later .
next , in the energy transmission phase with the remaining time @xmath100 , the et implements the oeb for wet based on the estimated mimo channel from the previous channel learning phase .
let the dominant eigenvector of the estimated mimo channel be denoted by @xmath101 .
then the et sets the transmit covariance matrix in the energy transmission phase as @xmath102 accordingly , the transferred energy to the er is expressed as @xmath103 by combing the channel learning and energy transmission phases , the net energy harvested by the er over one whole block is expressed as its totally harvested energy minus the total energy consumed for energy measurement and feedback , i.e. , @xmath104 accordingly , we define the practical energy beamforming gain achieved by the two - phase protocol as @xmath105 for typical wet applications with @xmath106 being sufficiently large , it follows that @xmath107 and @xmath108 .
furthermore , it can be assumed that the energy consumption @xmath109 and @xmath110 are negligible as compared to the totally harvested energy , i.e. , @xmath111 . versus @xmath100 between channel learning and energy transmission .
due to page limitation , we omit such discussion here and the interested readers can refer to @xcite for details .
] therefore , we approximate the practical energy beamforming gain as @xmath112 for convenience , in the sequel of this paper we use the approximate energy beamforming gain @xmath113 as the performance metric for proposed feedback and channel learning schemes .
note that when the channel estimation is sufficiently accurate , i.e. , @xmath114 , the energy beamforming gain @xmath113 approaches that by the oeb with perfect csi at the et @xmath115 in ( [ eqn : ebgain ] ) , i.e. , @xmath116 .
in this section , we present a general framework for designing the energy feedback at the er and the corresponding training signal and channel estimation at the et based on accpm .
accpm is well known as an efficient localization and cutting plane method for solving convex or quasi - convex optimization problems @xcite , with the objective of finding one feasible point in a convex target set , where the target set can be the set of optimal solutions to the problem of interest .
the implementation of accpm is based on the use of cutting planes , each of which is a hyperplane that separates some of the current points from the points in the target set .
based on the returned cutting planes in each iteration , the target set can be localized via a sequence of convex working sets with reduced size . to implement accpm in our context
, we first define the target set for channel learning as follows .
note that it follows from ( [ eqn : oeb ] ) that the implementation of oeb at the et in the energy transmission phase only depends on @xmath101 , which is the estimate of the dominant eigenvector of the matrix @xmath117 . for convenience , in the sequel
we denote @xmath51 as the mimo channel to be estimated , instead of the actual mimo channel @xmath118 . to obtain @xmath101
, it suffices for the et to estimate any positively scaled matrix of @xmath51 in the channel learning phase . without loss of generality , we focus on learning the normalized mimo channel given by @xmath119 with @xmath120 . accordingly , the target set of our interest is defined as @xmath121 .
next , we explain the principle for designing the feedback @xmath122 , and show that they in fact play the role of specifying the cutting planes in accpm for localizing @xmath123 . to facilitate the design of @xmath122
, we first set the transmit covariance matrix at the et in the first interval @xmath124 to be @xmath125 ( i.e. , isotropic transmission ) , based on which the corresponding harvested energy amount measured by the er is @xmath126 . by using @xmath127 as a reference energy level ,
we then normalize the harvested energy measurement in each interval @xmath128 as @xmath129 note that for interval @xmath124 , we have @xmath130 ; as a result , the feedback over this interval @xmath131 will not provide any information on @xmath132 and thus can be ignored .
this assumption is made specifically for our proposed channel learning algorithms , while in general it may not be optimal for other designs .
since @xmath133 is a real number but @xmath93 is of @xmath2 bits only , each feedback @xmath93 can not completely convey @xmath134 to the et at interval @xmath85 ; therefore , @xmath93 should be designed by properly encoding the present and past energy measurements , i.e. , @xmath135 , at each interval @xmath136 .
the optimal encoding scheme subject to @xmath2 feedback bits per interval is a difficult problem to solve ; thus we will later propose two suboptimal designs for efficient practical implementation in sections [ sec : quantization : feedback ] and [ sec : comparison : feedback ] , respectively .
for each proposed feedback scheme , a set of new linear inequalities ( or cutting planes in accpm ) on @xmath132 are obtained from @xmath93 at each interval @xmath137 , which are generally expressed as @xmath138 where @xmath139 denotes the number of inequalities / cutting planes obtained at interval @xmath85 , and @xmath140 and @xmath141 are parameters depending on the design of transmit covariance matrix @xmath142 s and feedback @xmath93 s in each scheme ( to be specified later ) .
each of the @xmath143 cutting planes in ( [ eqn : general : cut ] ) ensures that the mimo channel @xmath144 should lie in the half space of @xmath145 now , based on the returned cutting planes in ( [ eqn : general : cut ] ) ,
we are ready to obtain a sequence of working sets to localize the target set @xmath123 .
since @xmath146 and @xmath120 are _ a priori _ known by the et , we have the initial working set as @xmath147 , with @xmath148 .
furthermore , for the first interval @xmath124 , we have the working set @xmath149 , since @xmath150 is a constant and @xmath131 does not contain any information on @xmath132 . then , for the subsequent intervals @xmath136 , the working set can be updated as @xmath151 , or more explicitly , @xmath152 it is evident that @xmath153 .
therefore , the sequence of obtained working set @xmath154 s shrink in size and eventually converge to @xmath123 as @xmath155 . here , the convergence of our proposed accpm based channel learning can be proved , similarly as that given in @xcite for the special case of @xmath1 . for brevity , we omit the proof in this paper .
= 1 with accpm : ( a ) the working set @xmath156 containing the mimo channel @xmath132 to be learned ; ( b ) a neutral cutting plane that passes through @xmath157 , which is the analytic center of @xmath156 , to cut nearly a half space of @xmath156 , with @xmath158 ; ( c ) a deep cutting plane that locates between @xmath132 and @xmath157 to cut a more substantial sub - space of @xmath156 , with @xmath159.,title="fig:",width=302 ] finally , to complete the accpm based channel learning , it remains to design the cutting planes in ( [ eqn : general : cut ] ) in terms of @xmath140 and @xmath141 , @xmath160 , for each interval @xmath136 via adjusting the transmit covariance matrix @xmath142 .
consider a feedback interval @xmath85 with the given working set @xmath156 as shown in fig .
[ fig : framework : accpm](a ) . then , based on the principle of accpm @xcite , a desired cutting plane should be _ neutral _ or _
deep_. a neutral cutting plane should pass through the so - called analytic center of @xmath156 , denoted by @xmath157 , so that nearly a half portion of @xmath156 can be `` cut '' or eliminated , as shown in fig .
[ fig : framework : accpm](b ) ; while a deep cutting plane should locate between the analytic center @xmath157 and the mimo channel @xmath132 to be learned , such that a more substantial sub - space of @xmath156 can be cut as compared to a neural cutting plane , as shown in fig .
[ fig : framework : accpm](c ) . here , for each convex working set @xmath161 , its analytic center is defined as @xmath162 note that the minimization problem in ( [ eqn : accpm : general ] ) can be shown to be convex @xcite , and thus is solvable by standard convex optimization techniques such as cvx @xcite .
also note that we have @xmath163 from ( [ eqn : accpm : general ] ) . as a result , in order for the @xmath164th cutting plane to be neutral or deep ( see fig .
[ fig : framework : accpm ] ) , we design the transmit covariance matrix @xmath142 to ensure @xmath165 note that when @xmath166 , it may be difficult to guarantee all the @xmath143 cutting planes to be neutral or deep . in such cases , we can select any one of them to be the neutral / deep cutting plane to satisfy ( [ eqn : cut : general ] ) . the detailed design of such @xmath142 s will be explained later for each specific feedback scheme to be introduced . in summary ,
a generic algorithm for channel learning based on accpm is described as algorithm 1 in table i. note that at the last interval @xmath80 , the et adopts the analytic center @xmath167 as the final estimate of the mimo channel in the channel learning phase , given by @xmath168 . under the above general design framework with accpm , in the next two sections , we present two specific feedback schemes for channel learning , respectively . 1 ) * initialization : * set @xmath169 and @xmath170 . + 2 ) * repeat : * * @xmath171 ; * the et transmits with @xmath142 ; * the er determines @xmath93 ( if @xmath172 ) based on the normalized energy measurements , @xmath173 in ( [ eqn : barq ] ) , and then sends @xmath93 back to the et ; * the et extracts a set of @xmath143 cutting planes given in ( [ eqn : general : cut ] ) based on @xmath93 ( if @xmath172 ) , and computes the analytic center @xmath174 as in ( [ eqn : accpm : general ] ) ; * the et designs @xmath175 to ensure @xmath176 for some @xmath160 .
\3 ) * until * @xmath177 .
+ 4 ) the et obtains the channel estimate as @xmath178 . +
in this section , we propose an energy quantization based feedback scheme , where at each interval @xmath85 the er obtains the feedback @xmath93 by quantizing its normalized energy measurement @xmath134 at the current interval into @xmath2 bits .
for this scheme , we denote the feedback at the er as @xmath179 , the transmit covariance matrices at the et as @xmath180 , the updated sequence of analytic centers as @xmath181 , and the estimated mimo channel as @xmath182 .
first , we present the feedback design for @xmath179 at the er by quantizing the normalized energy measurements @xmath183 . among various quantization methods @xcite , we adopt a simple uniform scalar quantizer that is of low complexity for energy - efficient implementation at the er . consider an interval @xmath137 , and assume that the normalized energy measurement @xmath134 is within the unit interval @xmath184 $ ] . note that if an overflow ( i.e. , @xmath185 ) occurs , then we set @xmath186 . to minimize the probability of overflow , we should properly set the transmit power at the et as will be specified later ( see ( [ neutral:1 ] ) ) .
then , the uniform quantizer maps @xmath134 into a discrete variable that belongs to a finite set , given by @xmath187 , where there are in total @xmath188 quantization levels resulted from the @xmath2 feedback bits , and @xmath189 denotes the quantization step size . as a result , the quantized value is expressed as @xmath190 where @xmath191 denotes the minimum integer that is no smaller than @xmath23 .
accordingly , by encoding the quantized value @xmath192 as a @xmath2-bits digital codeword , the feedback @xmath193 can be obtained , which is then sent back to the et . after receiving the feedback @xmath193 , @xmath136
, the et has the knowledge of the quantized value @xmath192 , and accordingly knows that the normalized energy measurement @xmath134 should lie between two thresholds @xmath194 and @xmath195 , i.e. , , then the first inequality in ( [ eqn : inequations:1 ] ) becomes @xmath196 , which is known _ a priori _ and thus is redundant . on the other hand , if @xmath197 , then the second inequality in ( [ eqn : inequations:1 ] ) becomes @xmath198 , which , however , may be unreliable due to the possibility of overflow . in the above two cases , the corresponding inequalities should be discarded .
] @xmath199 by using ( [ eqn : inequations:1 ] ) together with ( [ eqn : barq ] ) , it follows that @xmath200 the two inequalities in ( [ eqn : inequations:2:1 ] ) and ( [ eqn : inequations:2 ] ) correspond to @xmath201 cutting planes in the general expression ( [ eqn : general : cut ] ) to localize the mimo channel @xmath132 with @xmath202 , @xmath203 , @xmath204 , and @xmath205 .
based on the feedback @xmath193 s , the et obtains the working set for the interval @xmath81 similarly as in ( [ eqn : working : set ] ) by replacing the cutting planes in ( [ eqn : general : cut ] ) as those in ( [ eqn : inequations:2:1 ] ) and ( [ eqn : inequations:2 ] ) , i.e. , @xmath206 accordingly , the analytic center of @xmath207 is expressed as @xmath208 finally , to complete the energy quantization based scheme , we design the transmit covariance matrix @xmath209 at each interval @xmath137 to make one of the cutting planes in ( [ eqn : inequations:2:1 ] ) and ( [ eqn : inequations:2 ] ) to be neutral or deep with respect to @xmath210 . towards this end , we design @xmath209 to ensure that @xmath211 where @xmath212 corresponds to the threshold between the two consecutive quantization levels @xmath213 and @xmath214 . from ( [ neutral:1 ] ) , it follows that when @xmath215 , we have @xmath216 , and hence the cutting plane in ( [ eqn : inequations:2:1 ] ) is neutral or deep ; otherwise ( when @xmath217 ) , we have @xmath218 , and thus the cutting plane in ( [ eqn : inequations:2 ] ) is also neutral or deep .
thus , both cases satisfy the requirement given in ( [ eqn : cut : general ] ) for accpm . in particular , at each interval @xmath137 , we set the transmit covariance matrix as @xmath219 where @xmath220 is an @xmath221 complex matrix , each element of which follows a circularly symmetric complex gaussian ( cscg ) distribution with zero mean and unit variance , and to satisfy ( [ neutral:1 ] ) @xmath222 is given by @xmath223 here , the randomness of @xmath220 in @xmath209 helps make the cutting planes over different feedback intervals as diverse as possible , so that they are more effective in cutting .
furthermore , several random trials may be required to make sure that @xmath224 . in summary ,
the energy quantization based feedback and channel learning scheme is described as algorithm 2 in table [ table:2:1 ] . 1 ) * initialization : * set @xmath169 and @xmath225 .
+ 2 ) * repeat : * * @xmath171 ; * the et transmits with @xmath209 ; * the er determines @xmath193 ( if @xmath226 ) by encoding the quantized value @xmath192 in ( [ eqn : quantize ] ) as a @xmath2-bits digital codeword , and then sends @xmath193 back to the et ; * the et updates the working set as @xmath207 in ( [ eqn : working : set:1 ] ) , and computes the analytic center of @xmath207 as @xmath227 given in ( [ eqn : accpm ] ) ; * the et generates @xmath228 given in ( [ eqn : s : lsm ] ) .
\3 ) * until * @xmath177 .
+ 4 ) the et obtains the channel estimate as @xmath229 . + it is interesting to discuss the energy quantization based scheme in the extreme case of unlimited number of feedback bits per interval , i.e. , @xmath231 . in this ideal case
, we have @xmath232 , and the two inequalities in ( [ eqn : inequations:2:1 ] ) and ( [ eqn : inequations:2 ] ) can be combined into one single linear equality as @xmath233 as a consequence , the computation of the analytic center in ( [ eqn : accpm ] ) at the interval @xmath234 can be re - expressed as @xmath235 to draw insights from problem ( [ eqn : accpm : b ] ) , we define a vector operation @xmath236 as follows . for any complex hermitian matrix @xmath237 that contains @xmath238 independent real elements , the vector operation @xmath236 maps @xmath239 to a real vector @xmath240 , where all elements of @xmath241 are independent from each other , and @xmath242 for any given complex hermitian matrix @xmath243 . and
the real vector @xmath244 , @xmath245 , can be realized as follows .
the first @xmath21 elements of @xmath246 consist of the diagonal elements of @xmath239 ( that are real ) , i.e. , @xmath247_{aa}$ ] s , @xmath248 , where @xmath247_{ab}$ ] denotes the element in the @xmath249th row and @xmath250th column of @xmath239 ; the next @xmath251 elements of @xmath246 are composed of the ( scaled ) real part of the upper ( or lower ) off - diagonal elements of @xmath239 , i.e. , @xmath252_{ab } + [ { \mbox{\boldmath{$ x $ } } } ] _ { ba}}{\sqrt{2}}$ ] s , @xmath253 ; and the last @xmath251 elements of @xmath246 correspond to the ( scaled ) imaginary part of the lower off - diagonal elements of @xmath239 , i.e. , @xmath254_{ab } - [ { \mbox{\boldmath{$ x $ } } } ] _ { ba}}{\sqrt{2}}$ ] s , @xmath253 . ] therefore , by defining @xmath255 and @xmath256 , we can rewrite the @xmath85 linear equality constraints in problem ( [ eqn : accpm : b ] ) as @xmath257 which correspond to a set of @xmath85 linear equations with @xmath258 real unknowns . based on ( [ eqn:4 ] )
, we discuss the solution @xmath259 in ( [ eqn : accpm : b ] ) by considering the following two cases with @xmath260 and @xmath261 , respectively .
* in the case with @xmath260 , the number of linear equations in ( [ eqn:4 ] ) is no smaller than that of the real unknowns . in this case ,
as long as the vector @xmath262 s are linearly independent such that @xmath263 with @xmath264^t \in \mathbb{r}^{n \times m_t^2}$ ] , the solution of @xmath265 to ( [ eqn:4 ] ) is unique , and so is the optimal solution @xmath259 in ( [ eqn : accpm : b ] ) due to the one - to - one mapping of @xmath236 . as a result
, the obtained @xmath259 is indeed the exact mimo channel @xmath132 to be learned , i.e. , @xmath266 . * in the case of @xmath261 ,
the set of equations in ( [ eqn:4 ] ) corresponds to an underdetermined system of linear equations . in this case , the solution of @xmath265 to ( [ eqn:4 ] ) is not unique in general , and so is the obtained @xmath259 in ( [ eqn : accpm : b ] ) .
as a result , @xmath259 may not be the exact mimo channel @xmath132 , i.e. , channel estimation error can occur . from the above discussion
, it is evident that when @xmath231 , a minimum number of @xmath258 feedback intervals are sufficient for the et to obtain an exact estimate of the mimo channel with the energy quantization based scheme .
in this section , we propose an alternative feedback design so - called energy comparison , where the @xmath2 feedback bits per interval indicate the increase or decrease of the harvested energy at the er in the current interval as compared to those in the previous @xmath2 intervals , respectively .
for this scheme , we denote the feedback at the er as @xmath267 , the transmit covariance matrices at the et as @xmath268 , the updated sequence of analytic centers as @xmath269 , and the estimated mimo channel as @xmath270 . in the energy comparison based scheme , the er designs each of the @xmath2 bits in @xmath271 separately .
consider a particular feedback interval @xmath136 and let the @xmath250th bit of @xmath271 be denoted by @xmath272 , @xmath273 .
then we use @xmath272 to indicate the increase or decrease of the normalized harvested energy @xmath134 by the er in the current interval @xmath85 as compared to @xmath274 in the @xmath250th previous interval , @xmath273 .
more explicitly , we have @xmath275 here , we assume @xmath276 and @xmath277 for any @xmath85 and @xmath250 with @xmath278 for notational convenience . by combining the obtained @xmath279 as @xmath271 ,
the er sends it back to the et . after receiving the feedback @xmath271 , at each interval @xmath136
, the et extracts the following @xmath280 linear inequalities on the mimo channel @xmath132 based on ( [ eqn : onebitfeedback ] ) as well as ( [ eqn : barq ] ) . in ( [ eqn : comfb ] ) , @xmath281 always holds due to @xmath282 . in this case
, the inequality due to @xmath281 becomes @xmath283 , which is trivial and does not provide any information on the mimo channel @xmath146 ( due to @xmath284 ) .
therefore , only @xmath285 s with @xmath286 are used for generating inequalities for channel learning . ]
@xmath287 which correspond to @xmath288 cutting planes in the general expression ( [ eqn : general : cut ] ) to localize the mimo channel with @xmath289 and @xmath290 . based on the feedback @xmath271 s , the et can obtain the working set for the interval @xmath81 similarly as in ( [ eqn : working : set ] ) by replacing the cutting planes in ( [ eqn : general : cut ] ) as those in ( [ eqn : comfb ] ) : @xmath291 accordingly , the analytic center of @xmath292 is expressed as @xmath293 finally , we design the transmit covariance matrix @xmath294 for the interval @xmath137 to make the cutting planes in ( [ eqn : comfb ] ) neutral with respect to @xmath295 ( thus , satisfying ( [ eqn : cut : general ] ) for accpm ) , i.e. , @xmath296 where the above equalities hold regardless of the sign of @xmath285 . however , if @xmath297 , then it is infeasible to find such an @xmath294 to satisfy all these equations in ( [ eqn : neutral ] ) at the same time . to overcome this difficulty , we propose to only ensure the neutral cutting plane for @xmath298 , i.e. , @xmath299 . towards this end
, we set @xmath300 here , @xmath301 is a hermitian probing matrix , which is neither positive nor negative semi - definite in general and satisfies @xmath302 accordingly , the @xmath303 cutting planes in ( [ eqn : comfb ] ) can be re - expressed as @xmath304 to find such a @xmath305 to satisfy ( [ eqn : delta ] ) for the @xmath85th interval , we express @xmath306 and @xmath307 , where @xmath308 . due to the one - to - one mapping of @xmath309 ,
finding @xmath305 is equivalent to finding @xmath310 that is orthogonal to @xmath311 . define a projection matrix @xmath312
. then we can express @xmath313 , where @xmath314 satisfies @xmath315 and @xmath316 .
thus , @xmath317 can be any vector in the subspace spanned by @xmath318 .
specifically , we set @xmath319 where @xmath320 is a randomly generated cscg vector with zero - mean and identity covariance matrix ( generally scaled by a certain positive constant ; see footnote 12 ) in order to make @xmath317 independently drawn from the subspace . with the obtained @xmath317 , we have @xmath321 , in general contains both positive and negative eigenvalues . as a result , the update in ( [ eqn : com : c_n ] ) may not necessarily yield an @xmath142 that satisfies both @xmath322 and @xmath323 .
nevertheless , by setting @xmath324 to be sufficiently smaller than @xmath325 , we can always find a @xmath326 and its resulting @xmath142 satisfying the above two conditions with only a few random trials . in this paper , we choose @xmath327 .
] where @xmath328 denotes the inverse operation of @xmath236 .
accordingly , @xmath142 that satisfies the equation in ( [ eqn : neutral ] ) for @xmath298 is obtained . in summary ,
the energy comparison based feedback and channel learning scheme is described as algorithm 3 in table [ table:3 ] . 1 ) * initialization : * set @xmath169 , and @xmath329 .
+ 2 ) * repeat : * * @xmath171 ; * the et transmits with @xmath294 ; * the er sends back @xmath330 ( if @xmath226 ) to the et with @xmath331 ( or @xmath332 ) if @xmath333 ( otherwise ) , @xmath273 ; * the et updates the working set as @xmath292 using ( [ eqn : working : set:2 ] ) , and computes the analytic center of @xmath292 as @xmath334 given in ( [ eqn : compar ] ) ; * the et generates @xmath335 , and updates @xmath336 .
\3 ) * until * @xmath177 .
+ 4 ) the et estimates the mimo channel as @xmath337 . + in this subsection , we compare the energy quantization versus energy comparison based feedback and channel learning schemes via geometric interpretations on their cutting planes at each feedback interval @xmath137 . for the ease of comparison , we assume that both schemes have the same working set @xmath338 and the same analytic center @xmath339 at interval @xmath340 , as shown in figs .
[ fig : comparison:1 ] and [ fig : comparison:2 ] , for the cases of large @xmath2 and small @xmath2 values , respectively .
first , we compare graphically the cutting planes returned by the two schemes . for the case of energy quantization , there are two cutting planes obtained at each interval @xmath85 as specified in ( [ eqn : inequations:2:1 ] ) and ( [ eqn : inequations:2 ] ) .
it is observed in figs .
[ fig : comparison:1](a ) and [ fig : comparison:2](a ) that the two cutting planes are parallel to each other , since the gradient of the left - hand - side ( lhs ) in ( [ eqn : inequations:2:1 ] ) is just the opposite of that in ( [ eqn : inequations:2 ] ) .
it is also observed that only one of the two cutting planes is deep ( e.g. , see fig .
[ fig : comparison:1](a ) ) or neutral ( e.g. , see fig .
[ fig : comparison:2](a ) ) with respect to the analytic center @xmath341 of @xmath156 , while the other is neither deep nor neutral .
this is consistent with the training signal design in ( [ neutral:1 ] ) .
= 1 .,title="fig:",width=302 ] in contrast , for the case of energy comparison , there are a total of @xmath303 cutting planes in each interval @xmath85 as specified in ( [ eqn : cuttingplanes : comparison ] ) .
it is observed in figs .
[ fig : comparison:1](b ) and [ fig : comparison:2](b ) that all these cutting planes pass through the common point of @xmath13 , while among them , only the neutral cutting plane corresponding to @xmath298 ( i.e. , @xmath342 ) passes through the analytic center @xmath295 of @xmath156 .
furthermore , the non - neutral cutting plane with @xmath343 ( i.e. , @xmath344 ) is observed to locate between the neutral cutting plane in the current interval @xmath85 with @xmath298 ( i.e. , @xmath342 ) and that in the previous interval @xmath340 with @xmath298 ( i.e. , @xmath345 , which passes through the analytic center @xmath346 of @xmath347 ) , since the gradient of the lhs in the former inequality is a linear combination of those in the latter two .
note that the point @xmath346 always locates on the boundary of @xmath156 ; as a result , the cutting plane with @xmath348 is very likely to cross the interior of @xmath156 , making it efficiently eliminate a sizable portion of @xmath156 .
similar properties also hold for other non - neutral cutting planes with @xmath349 .
= 1 .,title="fig:",width=302 ] next , with the above geometric interpretations , we compare the effectiveness of the two feedback schemes for channel learning by considering three cases with different values of @xmath2 . *
first , consider the case of large @xmath2 as shown in fig .
[ fig : comparison:1 ] . for energy quantization ,
the gap between the two cutting planes is small ( particularly , it is zero when @xmath350 ) , and as a result they can cut a significant portion of @xmath156 .
therefore , energy quantization is more effective than energy comparison in this case .
* next , consider the case of small @xmath2 ( but with @xmath351 ) as depicted in fig .
[ fig : comparison:2 ] . for energy quantization ,
the gap between the two cutting planes becomes significantly larger as compared to the case of large @xmath2 in fig .
[ fig : comparison:1 ] . as a result ,
the non - neutral cutting plane is likely to be less efficient or even redundant .
in contrast , for energy comparison , the non - neutral cutting planes will still cross the interior of @xmath156 , which helps cut the volume of @xmath156 effectively even with a small @xmath2 value . as a result ,
the cutting planes of energy comparison are generally more effective than those of energy quantization in this case . * finally , consider the special case of @xmath1 . in this case , both schemes can only return one valid cutting plane per interval .
intuitively , the two schemes should have similar channel learning performance ; however , as will be shown by the numerical results in section [ sec : numerical : results ] , energy comparison is still more effective than energy quantization in this case .
in the previous sections , we have proposed a general channel learning design framework based on accpm as well as two specific feedback schemes for the et to learn the csi , where at each feedback interval the et needs to solve an optimization problem ( see ( [ eqn : accpm : general ] ) ) for obtaining the analytic center of the updated working set . as the number of feedback intervals @xmath85 increases and that of added cutting planes becomes large , the complexity for solving such problems also increases significantly . to solve this issue ,
in this section we propose an efficient complexity reduction method by pruning irrelevant cutting planes at each feedback interval @xmath85 after obtaining the analytic center of @xmath161 ( i.e. , by adding a pruning procedure between steps 2-d ) and 2-e ) in algorithms 1 - 3 ) . for brevity , we discuss for the general design framework in section [ sec : framework ] only , while similar procedures can be applied for the two specific schemes in sections [ sec : quantization : feedback ] and [ sec : comparison : feedback ] .
the complexity reduction method is motivated by the example illustrated in fig .
[ fig : pruning ] , where various cutting planes returned at intervals @xmath352 establish a working set @xmath154 for localizing the target set @xmath353 .
it is observed that among these cutting planes , some of them are important active constraints to bound @xmath154 ( see the solid lines in fig .
[ fig : pruning ] ) , while the others are less relevant or even redundant ( see the dashed and dotted lines in fig . [
fig : pruning ] ) .
therefore , by pruning such irrelevant cutting planes , we could reduce the computation complexity for solving problem ( [ eqn : accpm : general ] ) without degrading the channel learning performance . in the following
, we first rank the irrelevance of the cutting planes .
= 1 ( shown shaded ) characterized by various cutting planes ( depicted as lines ) . here ,
the solid lines represent most important cutting planes for bounding @xmath161 ; the dashed line shows a less important cutting plane ( though still active for bounding @xmath161 ) ; and the dotted line is a redundant cutting plane.,title="fig:",width=188 ] consider a particular feedback interval @xmath136 and suppose that there are @xmath354 cutting planes at hand , which form a polyhedron as @xmath355 as compared to the working set @xmath161 in ( [ eqn : working : set ] ) , we have omitted @xmath356 and @xmath357 in ( [ eqn : polyhedron ] ) to focus on studying the cutting planes only . for simplicity
, we use @xmath358 ( i.e. , the analytic center of @xmath161 in ( [ eqn : working : set ] ) ) as the ( approximate ) analytic center of the polyhedron in ( [ eqn : polyhedron ] ) , since the constraints @xmath356 and @xmath357 become less likely to be active when the number of cutting planes @xmath354 becomes large .
define @xmath359 then , the ellipsoid @xmath360 lies inside the polyhedron in ( [ eqn : polyhedron ] ) , while the ellipsoid @xmath361 which is @xmath362 scaled by a factor of @xmath354 over its center , contains the polyhedron in ( [ eqn : polyhedron ] ) ( * ? ? ?
* chapter 8.5 ) .
therefore , the ellipsoid @xmath362 at least grossly approximates the shape of the polyhedron in ( [ eqn : polyhedron ] ) , and indicates the following irrelevance measure @xcite @xmath363 for the cutting plane @xmath364 .
here , larger @xmath365 means that the corresponding cutting plane is more likely to be irrelevant .
next , we prune irrelevant cutting planes based on the above irrelevance measure @xmath366 .
we simply drop the cutting planes with largest values of @xmath365 , and keep the cutting planes with smaller values . in particular , the number of kept cutting planes is fixed as @xmath367 , where @xmath368 is a design parameter controlling the trade - off between the channel learning accuracy and computation complexity ( as will be shown in more detail via numerical results in section [ sec : pruning ] ) . with fixed @xmath368 ,
the computational complexity per interval for solving problem ( [ eqn : accpm : general ] ) does not grow as @xmath85 increases .
in this section , we provide numerical results to evaluate the performance of our proposed channel learning schemes .
we consider a point - to - point mimo wet system with one er located at a distance of 5 meters from the et , where the average path loss from the et to the er is 40 db .
for the considered short transmission distance , the los signal is dominant , and thus the rician fading is used to model the channel from the et to the er . specifically , we assume @xmath369 where @xmath370 is the los deterministic component , @xmath371 denotes the non - los rayleigh fading component with each element being an independent cscg random variable with zero mean and variance of @xmath372 ( to be consistent with the assumed average power attenuation of @xmath373 db ) , and @xmath374 is the rician factor set to be @xmath375 db . for the los component
, we use the far - field uniform linear antenna array model with each row of @xmath376 expressed as @xmath377 $ ] with @xmath378 , where @xmath379 is the spacing between two adjacent antenna elements at the et , @xmath380 is the carrier wavelength , and @xmath381 is the direction of the er from the et .
we set @xmath382 and @xmath383 .
furthermore , we set the number of transmit antennas at the et as @xmath384 , the number of receive antennas at the er as @xmath385 , and the maximum transmit sum - power as @xmath386 dbm ( @xmath332 watt ) .
= 1 with @xmath1 and @xmath387.,title="fig:",width=302 ] to start with , we compare the channel learning performance of the energy quantization and energy comparison based feedback schemes . figs .
[ fig : channellearning1 ] and [ fig : channellearning2 ] plot the normalized error of estimated matrix norm , i.e. , @xmath388 , versus the number of feedback intervals @xmath80 , under different numbers of feedback bits per interval , @xmath2 .
when @xmath1 , it is observed in fig .
[ fig : channellearning1 ] that energy comparison achieves a lower normalized error than energy quantization . when @xmath351 , it is observed in figs .
[ fig : channellearning1 ] and [ fig : channellearning2 ] that energy quantization outperforms energy comparison when @xmath80 is below a certain threshold ( e.g. , @xmath389 for @xmath387 , and @xmath390 for @xmath391 ) , while the opposite is true when @xmath80 becomes larger than the threshold .
this result can be explained as follows based on the geometric interpretations of their different cutting planes ( for a given interval @xmath85 ) in section [ sec : comparsion2 ] .
when the number of feedback intervals @xmath85 is small , the working set @xmath161 is large and energy quantization is more effective in space cutting with a fixed @xmath2 ; however , when @xmath85 is large , @xmath161 becomes small and energy comparison is more effective due to multiple cutting planes with the same @xmath2 value .
furthermore , when @xmath350 , it is observed in fig .
[ fig : channellearning2 ] that energy quantization obtains an exact estimate of the mimo channel with the normalized error falling sharply to zero after @xmath392 intervals .
this result is consistent with our discussion in section [ sec : perfect : feedback ] .
= 1 with @xmath393 , @xmath391 , and @xmath231.,title="fig:",width=302 ] = 1 = 1 with @xmath1 and @xmath387.,title="fig:",width=302 ] = 1 with @xmath393 , @xmath391 , and @xmath231.,title="fig:",width=302 ] next , we evaluate the energy beamforming performance with estimated mimo channels by the two schemes , as compared to a reference scheme named random beamforming . similar to our proposed accpm based channel learning ,
the random beamforming scheme implements a two - phase transmission protocol for channel learning and energy transmission , respectively , as shown in fig .
[ fig : twophase : ref ] .
however , the channel learning phase in this case is divided into @xmath394 intervals , with the length of @xmath87 for the first @xmath395 intervals and @xmath396 for the last one . in the @xmath85th interval , @xmath397 ,
the et randomly generates a beamforming vector @xmath398 with @xmath399 and transmits with the covariance matrix @xmath400 . at the same time , the er measures the harvested energy @xmath401 . in the @xmath402th interval ,
the er compares the measured energy amounts @xmath403 to obtain the maximum value , and sends its index @xmath404 back to the et .
hence , the number of feedback bits in the @xmath402th interval is given by @xmath405 .
after receiving the feedback information , in the second energy transmission phase , the et transmits with the energy beamforming vector @xmath406 in the @xmath407th interval of the channel learning phase . for a fair comparison
, we set the length of the channel learning phase in the random beamforming scheme to be same as that in our proposed accpm based channel learning scheme , i.e. , @xmath408 .
furthermore , to obtain an accurate measurement result , the time consumed is in general larger than that for the feedback in each interval ; hence , we assume @xmath409 and @xmath410 in the simulation by ignoring the time difference in sending back different number of bits .
thus , we have @xmath411 . figs .
[ fig : energbeamgain1 ] and [ fig : energbeamgain2 ] depict the achieved energy beamforming gain @xmath412 given by ( [ eqn : chieb ] ) in db scale versus the number of feedback intervals @xmath80 under different values of @xmath2 .
it is observed that when @xmath80 is small ( e.g. , @xmath413 ) , energy comparison ( energy quantization ) achieves a higher energy beamforming gain in the case of @xmath1 ( @xmath393 and @xmath391 ) , while the two schemes have a similar energy beamforming gain when @xmath387 .
this result is consistent with the channel learning performance comparison in figs .
[ fig : channellearning1 ] and [ fig : channellearning2 ] .
in addition , when @xmath80 becomes large ( e.g. , @xmath414 ) , both schemes with different values of @xmath2 are observed to achieve energy beamforming gains close to the optimal one by oeb in the perfect csi case ( i.e. , @xmath115 defined in ( [ eqn : ebgain ] ) ) . based on this observation together with those in figs .
[ fig : channellearning1 ] and [ fig : channellearning2 ] , it is evident that a relatively rough channel estimation is sufficient to achieve a close - to - optimal energy beamforming gain for the mimo wet system .
furthermore , it is observed from figs .
[ fig : energbeamgain1 ] and [ fig : energbeamgain2 ] that the random beamforming scheme performs much worse than our proposed accpm based channel learning schemes .
also note that unlike our proposed algorithms , the random beamforming scheme can only estimate the optimal beamforming vector , but can not be used to estimate the exact mimo channel .
= 1 , under different values of @xmath80.,title="fig:",width=302 ] furthermore , it is interesting to consider a practical scenario where the number of feedback intervals @xmath80 is kept small .
[ fig : energbeamgain3 ] shows the energy beamforming gain versus the number of feedback bits per interval @xmath2 , with @xmath415 and @xmath416 .
it is observed that as @xmath2 increases , the energy beamforming gain by energy quantization improves significantly , while that by energy comparison almost remains unchanged .
this shows that energy quantization is more effective in the regime of large @xmath2 but limited @xmath80 .
to summarize , the above performance comparison provides useful insights on the selection between the energy quantization and energy comparison schemes in practice . for mimo wet systems requiring relatively coarse channel estimation ,
energy quantization is more effective when the number of feedback bits per interval @xmath2 is large , while energy comparison is more suitable when @xmath2 is small .
however , for other applications that require more accurate channel estimation ( e.g. , communication systems such as cognitive radio networks @xcite ) , energy comparison could be more effective even under large values of @xmath2 , provided that the number of feedback intervals @xmath80 is sufficiently large .
furthermore , we evaluate the performance of the complexity reduction method via cutting planes pruning , in terms of both normalized error of estimated matrix norm and achieved energy beamforming gain . due to space limitation
, we only consider the energy quantization based scheme , while similar results have been observed for energy comparison and thus are omitted .
we set @xmath387 in this subsection .
= 1 for energy quantization with cutting planes pruning.,title="fig:",width=302 ] fig .
[ fig : pruning1 ] shows the normalized error of estimated matrix norm versus the number of feedback intervals @xmath80 under different number of kept cutting planes @xmath368 .
it is observed that the cases with @xmath417 and @xmath418 achieve similar normalized errors of estimated matrix norm as that without pruning , while the case with @xmath419 results in noticeable performance degradation .
= 1 for energy quantization with cutting planes pruning.,title="fig:",width=302 ] fig .
[ fig : pruning2 ] depicts the energy beamforming gain versus @xmath80 .
it is observed that all the cases with @xmath420 , @xmath417 , and @xmath418 achieve similar energy beamforming gains as compared to that without pruning .
this demonstrates the effectiveness of the method of cutting planes pruning . in the above studies
, we have assumed that the energy measurement at the er is perfect and has no errors for the purpose of exposition . in the last subsection
, we investigate the effect of energy measurement errors ( e.g. due to the distortion induced by the energy harvesting circuits ) on the channel learning performances by simulation .
it is assumed that the measured energy amount at the er in each interval @xmath81 is @xmath421 , where @xmath98 is the exact harvested energy given in ( [ eqn1:measured ] ) and @xmath422 denotes the error in the energy meter reading which follows a uniform distribution on the interval @xmath423 $ ] with @xmath424 being a constant .
consider the general design framework for channel learning in section [ sec : framework ] . due to the energy measurement error , at each interval @xmath85 the cutting planes in ( [ eqn : general : cut ] ) may not hold in general and thus it may occur that @xmath425 . to resolve this issue
, we modify the cutting planes in ( [ eqn : general : cut ] ) as @xmath426 and accordingly obtain the relaxed working set at each interval @xmath81 as @xmath427 where @xmath428 are parameters designed for ensuring @xmath429 with improved accuracy .
specifically , we set @xmath430 as the optimal solution of @xmath431 to the following convex optimization problem to minimize the distortion to the cutting planes .
@xmath432 then the analytic center of the modified working set @xmath433 is given by @xmath434 as a result , the modified generic algorithm can be obtained similarly as algorithm 1 by replacing the cutting planes in ( [ eqn : general : cut ] ) by those in ( [ eqn : working : set : relaxed ] ) , and the analytic center @xmath435 by @xmath436 .
note that the energy quantization and energy comparison based channel learning schemes should be accordingly modified as above in the case with energy measurement errors .
= 1 under energy measurement errors.,title="fig:",width=302 ] = 1 under energy measurement errors.,title="fig:",width=302 ] fig .
[ fig : error1 ] shows the normalized error of estimated matrix norm versus
the number of feedback intervals @xmath80 subject to different values of energy measurement errors with @xmath387 .
it is observed that for both channel learning schemes , the measurement errors with @xmath437 and @xmath438 result in error floors for channel estimation .
in addition , when @xmath439 , energy comparison is observed to perform worse than energy quantization in the case with @xmath438 , which is in sharp contrast to the case without energy measurement errors ( i.e. , @xmath440 ) , where the reverse is true .
this shows that energy comparison is generally more sensitive to the energy measurement errors than energy quantization .
[ fig : error2 ] shows the resulting energy beamforming gain versus @xmath80 under different levels of energy measurement errors with @xmath387 .
it is observed that with @xmath437 and @xmath438 , the two channel learning schemes have similar energy beamforming gains as that without energy measurement errors ( i.e. , @xmath440 ) , despite the error floor phenomenon observed for the normalized channel matrix in fig .
[ fig : error1 ] .
this shows that in mimo wet systems , our proposed channel learning schemes ( with above robust design ) are immune to the energy measurement errors .
this paper studied a new form of energy measurement based feedback for channel estimation in mimo wet systems , by taking into account the energy and hardware limitations of practical ers .
we proposed a general channel learning design framework , by leveraging the technique of accpm in convex optimization . under this framework
, we developed two energy feedback schemes , namely energy quantization and energy comparison , for efficiently encoding the energy measurements at the er with limited feedback .
simulation results showed that energy quantization is generally more effective when the number of feedback bits per interval is large , while energy comparison is preferable with small number of feedback bits .
it is our hope that this paper provides new insights and will inspire future investigations on the practical design of channel learning for mimo wet systems or other wireless communication systems based on the signal energy / power feedback .
y. che , j. xu , l. duan , and r. zhang , `` multiantenna wireless powered communication with co - channel energy and information transfer , '' _ ieee commun . letters _
2266 - 2269 , dec . 2015 .
y. zeng and r. zhang , `` optimized training for net energy maximization in multi - antenna wireless energy transfer over frequency - selective channel , '' _ ieee trans .
63 , no . 6 , pp .
2360 - 2373 , jun .
g. yang , c. k. ho , r. zhang , and y. l. guan , `` throughput optimization for massive mimo systems powered by wireless energy transfer , '' _ ieee j. sel .
areas commun .
33 , no . 8 , pp .
1640 - 1650 , aug .
2015 .
d. j. love , r. w. heath jr .
, v. k. n. lau , d. gesbert , b. d. rao , and m. andrews , `` an overview of limited feedback in wireless communication systems , '' _ ieee j. sel .
areas commun .
1341 - 1365 , oct .
|
multi - antenna or multiple - input multiple - output ( mimo ) technique can significantly improve the efficiency of radio frequency ( rf ) signal enabled wireless energy transfer ( wet ) . to fully exploit the energy beamforming gain at the energy transmitter ( et ) ,
the knowledge of channel state information ( csi ) is essential , which , however , is difficult to be obtained in practice due to the energy and hardware limitation of the energy receiver ( er ) . to overcome this difficulty , under a point - to - point mimo wet setup , this paper proposes a general design framework for a new type of channel learning method based on the er s energy measurement feedback . specifically , the er measures and encodes the harvested energy levels over different training intervals into bits , and sends them to the et via a feedback link of limited rate . based on the energy - level feedback , the et adjusts transmit beamforming in subsequent training intervals and obtains refined estimates of the mimo channel by leveraging the technique of analytic center cutting plane method ( accpm ) in convex optimization . under this general design framework
, we further propose two specific feedback schemes based on energy quantization and energy comparison , where the feedback bits at each interval are generated at the er by quantizing the measured energy level at the current interval and comparing it with those in previous intervals , respectively .
numerical results are provided to compare the performance of the two feedback schemes .
it is shown that energy quantization performs better when the number of feedback bits per interval is large , while energy comparison is more effective vice versa .
wireless energy transfer ( wet ) , energy beamforming , channel learning , limited feedback , analytic center cutting plane method ( accpm )
. [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] [ section ]
|
the _ parameter space _ of complex degree @xmath0 polynomials is by definition the space of affine conjugacy classes of these polynomials .
equivalently , one can talk about the space of all _ monic central polynomials _ of degree @xmath0 , i.e. polynomials of the form @xmath1 .
any polynomial is affinely conjugate to a monic central polynomial .
an important set is the _ connectedness locus _ @xmath2 consisting of classes of all degree @xmath0 polynomials @xmath3 , whose julia sets @xmath4 ( equivalently , whose _ filled julia sets _
@xmath5 ) are connected .
general properties of the connectedness locus @xmath2 have been studied for quite some time .
for instance , it is known that @xmath2 is a compact cellular set in the parameter space of complex degree @xmath0 polynomials ( this was proven in @xcite in the cubic case and in @xcite for higher degrees , see also @xcite ; by definition , following m. brown @xcite , a subset of a euclidean space @xmath6 is _ cellular _ if its complement in the sphere @xmath7 is an open topological cell ) . for @xmath8
, a monic centered polynomial takes the form @xmath9 , and the parameter space of quadratic polynomials can be identified with the plane of complex parameters @xmath10 .
clearly , @xmath11 has a unique critical point @xmath12 and a unique critical value @xmath10 in @xmath13 .
thus , we can say that polynomials @xmath11 are parameterized by their critical values .
the quadratic connectedness locus is the famous _ mandelbrot set _ @xmath14 , identified with the set of complex numbers @xmath10 not escaping to infinity under iterations of the polynomial @xmath11 .
the mandelbrot set @xmath14 has a complicated self - similar structure .
the `` pinched disk '' model for @xmath14 is due to douady and thurston @xcite . to describe their approach to the problem of modeling @xmath14 , we first describe _
laminational _ models of polynomial julia sets ( we follow @xcite ) and then use them to interpret results of @xcite .
we assume basic knowledge of complex dynamics ( a detailed description is given later ) .
let @xmath15 be the unit circle in @xmath13 , consisting of all complex numbers of modulus one .
we write @xmath16 for the restriction of the map @xmath17 .
we identify @xmath15 with @xmath18 by the mapping taking an _ angle _
@xmath19 to the point @xmath20 . under this identification
, we have @xmath21 .
we will write @xmath22 for the open unit disk @xmath23 . given a complex polynomial @xmath3 , we let @xmath24 denote the set @xmath25 .
this set is called the _ basin of attraction of infinity _ of @xmath3 .
clearly , @xmath26 .
if the julia set @xmath4 is locally connected , then it is connected , and the riemann map @xmath27 can be continuously extended to a map @xmath28 .
this gives rise to a map @xmath29 , which semiconjugates @xmath30 with @xmath31 .
define an equivalence relation @xmath32 on @xmath15 so that @xmath33 if and only if @xmath34 . then @xmath35 and @xmath4 are homeomorphic , and the homeomorphism in question conjugates the map @xmath36 induced on @xmath35 by @xmath37 , and @xmath31 .
it is not hard to see that the convex hulls of @xmath32-classes are disjoint in @xmath38 .
a productive idea is to consider equivalences relations @xmath39 whose properties are similar to those of @xmath32 .
these properties will be stated precisely below .
such equivalence relations are called _ laminational equivalence relations ( of degree @xmath0 _ ; if @xmath8 they are said to be _ quadratic _ , and if @xmath40 they are said to be _
cubic_. the maps @xmath41 induced by @xmath37 are called _ topological polynomials _ of degree @xmath0 ; again , if @xmath8 they are called _ quadratic _ and if @xmath40 they are called _ cubic_. for brevity , in what follows we will talk about `` @xmath39-classes '' instead of `` classes of equivalence of @xmath39 '' . an important geometric interpretation of a laminational equivalence relation @xmath39 is as follows .
for any @xmath39-class @xmath42 , take its convex hull @xmath43 .
consider the edges of all such convex hulls ; add all points of @xmath15 to this collection of chords .
the obtained collection of ( possibly , degenerate ) chords in the unit disk is denoted by @xmath44 and is called a _ geodesic lamination ( generated by @xmath39)_. for brevity in what follows we sometimes write `` lamination '' instead of `` geodesic lamination '' . clearly , @xmath44 is a closed family of chords .
let @xmath45 denote the straight line segment connecting points @xmath46 , @xmath47 .
we will never use this notation for pairs of points not in @xmath15 . recall also that points in @xmath15
are identified with their angles .
thus , @xmath48 always means the chord of @xmath15 connecting the points with angles @xmath12 and @xmath49 ( not a half - radius of the unit disk originating at the center ) . for any chord @xmath50 in the closed unit disk @xmath38
set @xmath51 .
for any @xmath39-class @xmath42 and , more generally , for any closed set @xmath52 , we set @xmath53 .
recall the construction of douady and thurston .
suppose that a quadratic polynomial @xmath54 has locally connected julia set .
we will write @xmath55 for the convex hull of the @xmath56-class corresponding to the critical value @xmath10 .
a fundamental theorem of thurston is that @xmath57 implies that @xmath55 and @xmath58 are disjoint in @xmath38 .
consider the collection of all @xmath55 and take its closure .
the thus obtained collection of chords and inscribed polygons defines a geodesic lamination @xmath59 introduced by thurston in @xcite and called the _ quadratic minor lamination _ ; moreover , it induces an equivalence relation @xmath60 on @xmath15 @xcite .
the corresponding quotient space @xmath61 is a combinatorial model for the boundary of @xmath14 .
it is called the _ combinatorial mandelbrot set_. conjecturally , the combinatorial mandelbrot set is homeomorphic to the boundary of @xmath14 .
this conjecture is equivalent to the famous mlc conjecture : the mandelbrot set is locally connected .
when defining the combinatorial mandelbrot set , we used a partial association between parameter values @xmath10 and laminational equivalence relations @xmath56 . in order to talk about @xmath56
, we had to assume that @xmath62 was locally connected .
recall that a _
dendrite _ is a locally connected continuum that does not contain jordan curves .
recall also , that a map from a continuum to a continuum is called _ monotone _
if under this map point - preimages ( _ fibers _ ) are connected .
[ d : dendr - poly ] a complex polynomial @xmath3 is said to be _
dendritic _ if it has connected julia set and all cycles repelling .
a topological polynomial is said to be _
dendritic _ if its julia set is a dendrite . in that case
the corresponding laminational equivalence relation and the associated geodesic lamination are also said to be _
it is known that there are dendritic polynomials with non - locally connected julia sets .
nevertheless , by @xcite , for _ every _ dendritic polynomial @xmath3 of degree @xmath0 , there is a monotone semiconjugacy @xmath63 between @xmath64 and a certain topological polynomial @xmath36 such that the map @xmath63 is one - to - one on all periodic and pre - periodic points of @xmath3 .
moreover , by @xcite the map @xmath63 is unique and can be defined in a purely topological way . call a monotone map @xmath65 of a connected polynomial julia set @xmath66 onto a locally connected continuum @xmath67 the _ finest monotone map of @xmath4 onto a locally connected continuum _
if , for any monotone @xmath68 with @xmath69 locally connected , there is a monotone map @xmath70 with @xmath71 .
then it is proven in @xcite that for _ any polynomial _ the finest monotone map on a connected polynomial julia set semiconjugates @xmath31 to the corresponding topological polynomial @xmath36 on its topological julia set @xmath72 generated by the laminational equivalence relation possibly with infinite classes @xmath32 , and that in the dendritic case this semiconjugacy coincides with the map @xmath63 constructed by kiwi in @xcite .
clearly , this shows that @xmath63 is unique .
thus , @xmath3 gives rise to a corresponding laminational equivalence relation @xmath32 even if @xmath4 is not locally connected . if @xmath9 is a quadratic dendritic polynomial , then @xmath55 is defined , and is a finite - sided polygon inscribed into @xmath15 , or a chord , or a point . a parameter value @xmath10 is said to be _ quadratic dendritic _ if @xmath54 is dendritic .
the fundamental results of thurston @xcite imply , in particular , that @xmath55 and @xmath58 are either the same or disjoint , for all pairs @xmath10 , @xmath73 of dendritic parameter values .
moreover , the mapping @xmath74 is _ upper semi - continuous _ (
if a sequence of dendritic parameters @xmath75 converges to a dendritic parameter @xmath10 , then the limit set of the corresponding convex sets @xmath76 is a subset of @xmath55 ) .
we call @xmath55 the _ tag associated to @xmath10_. now , consider the union of all tags of quadratic dendritic polynomials .
this union is naturally partitioned into individual tags ( distinct tags are pairwise disjoint ! ) .
thus the space of tags can be equipped with the quotient space topology induced from the union of tags .
on the other hand , take the set of quadratic dendritic parameters .
each such parameter @xmath10 maps to the polygon @xmath55 , i.e. to the tag associated to @xmath10 .
thus each quadratic dendritic parameter maps to the corresponding point of the space of tags .
this provides for a combinatorial ( or laminational ) model for the set of quadratic dendritic polynomials ( or their parameters ) . in this paper
, we extend these results to cubic dendritic polynomials .
recall that monic centered quadratic polynomials are parameterized by their critical values . a combinatorial analog of this parameterization is the association between topological polynomials and their tags .
tags of quadratic topological polynomials are post - critical objects of the corresponding laminational equivalences .
monic centered cubic polynomials can be parameterized by a critical value and a co - critical point .
recall that the _ co - critical _ point @xmath77 of a cubic polynomial @xmath3 corresponding to a simple critical point @xmath78 of @xmath3 is defined as a point different from @xmath78 but having the same image under @xmath3 as @xmath78 . if @xmath78 is a multiple critical point of @xmath3
, then we set @xmath79 . in any case
we have @xmath80 .
let @xmath10 and @xmath0 be the two critical points of @xmath3 ( if @xmath3 has a multiple critical point , then @xmath81 ) .
set @xmath82 and @xmath83 .
assuming that @xmath3 is monic and central , we can parameterize @xmath3 by @xmath46 and @xmath84 : @xmath85 for @xmath3 in this form , we have @xmath86 , @xmath87 .
similarly to parameterizing cubic polynomials by pairs @xmath88 , we will use the so - called _ mixed tags _ to parameterize topological cubic dendritic polynomials . consider a cubic dendritic polynomial @xmath3 . by the above there
exists a laminational equivalence relation @xmath32 and a monotone semiconjugacy @xmath89 between @xmath90 and the topological polynomial @xmath36 . given a point @xmath91 , we associate with it the convex hull @xmath92 of the @xmath32-equivalence class represented by the point @xmath93 . if @xmath3 is fixed , we may write @xmath94 instead of @xmath92 . the set @xmath94 is a convex polygon with finitely many vertices , a chord , or a point ; it should be viewed as a combinatorial object corresponding to @xmath95 . for any points @xmath96 , the sets @xmath94 and @xmath97 either coincide or are disjoint . by definition , a _ @xmath98critically@xmath99 marked _
( cf @xcite ) cubic polynomial is a triple @xmath100 , where @xmath3 is a cubic polynomial with critical points @xmath10 and @xmath0 . if @xmath3 has only one ( double ) critical point , then @xmath81 , otherwise @xmath101 .
in particular , if @xmath101 , then the triple @xmath100 and the triple @xmath102 are viewed as two distinct critically marked cubic polynomials . when the order of the critical points is fixed , we will sometimes write @xmath3 instead of @xmath100 .
critically marked polynomials do not have to be dendritic ( in fact , the notion is used by milnor and poirier @xcite for hyperbolic polynomials , i.e. , in the situation diametrically opposite to that of dendritic polynomials ) .
however in this paper whenever we talk about critically marked polynomials we mean that they are dendritic .
let @xmath103 be the space of all critically marked cubic dendritic polynomials . with every marked dendritic polynomial @xmath100
, we associate the corresponding _ mixed tag _
@xmath104 here @xmath105 is the co - critical point corresponding to the critical point @xmath10 . a similar construction can be implemented for any cubic dendritic laminational equivalence relation @xmath39 .
let @xmath106 and @xmath107 denote the convex hulls of its critical classes .
then either @xmath108 is the unique critical @xmath39-class or @xmath109 are disjoint . the sets @xmath106 and @xmath107 are called _ critical objects _ of @xmath39 . by a _
@xmath98critically@xmath99 marked cubic laminational equivalence relation _ we mean a triple @xmath110 ; in that case we always assume that @xmath39 is dendritic .
if @xmath109 , then we define @xmath111 as the convex hull of the unique @xmath39-class that is distinct from the class @xmath112 but has the same @xmath113-image . if @xmath108 , then we set @xmath114 .
the set @xmath115 is called the _ co - critical set of @xmath106_. for a marked laminational equivalence relation @xmath110 , define its _ mixed tag _ as @xmath116 we endow the family of products of compact subsets of @xmath38 with the product topology on @xmath117 .
it is easy to see that the map @xmath118 is continuous as a map defined on a subset of @xmath117 . evidently , the map @xmath118 preserves inclusions .
the subscript @xmath119 in @xmath118 indicates that @xmath118 acts on marked laminational equivalence relations unlike the map @xmath120 that acts on polynomials .
these two maps are closely related though : for any marked dendritic cubic polynomial @xmath100 and the corresponding marked laminational equivalence relation @xmath121 , we have @xmath122 .
consider the collection of the sets @xmath123 over all @xmath124 .
by @xcite , for any dendritic laminational equivalence relation @xmath39 , there exists a dendritic complex polynomial @xmath3 with @xmath125 .
thus , equivalently , we can talk about the collection of mixed tags of all dendritic laminations @xmath44 . in theorem [ t : tagusc ]
we show that the mixed tags @xmath123 are pairwise disjoint or equal .
let us denote this collection of sets be @xmath126 ( for _ cubic mixed lamination of dendritic polynomials _ ) .
note , that @xmath126 can be viewed as ( non - closed ) `` lamination '' in @xmath127 whose elements are products of points , leaves or gaps .
one can consider @xmath126 as the higher - dimensional analog of thurston s @xmath59 restricted to dendritic polynomials .
theorem [ t : tagusc ] , in addition , establishes the fact that the collection of sets @xmath126 is upper semi - continuous .
let the _ union _ of all sets in @xmath126 be denoted by @xmath128 .
it follows that the quotient space of @xmath129 , obtained by collapsing all elements of @xmath126 to points , is a separable metric space which we denote by @xmath130 .
denote by @xmath131 the corresponding quotient map .
mixed tags of critically marked polynomials from @xmath103 are disjoint or coincide .
the map @xmath132 is continuous .
hence @xmath130 is a combinatorial model for @xmath133 .
this theorem can be viewed as a partial generalization of thurston s results @xcite to cubic polynomials .
it is also a first step towards defining a combinatorial model for @xmath134 .
lavaurs @xcite proved that @xmath134 is not locally connected .
epstein and yampolsky @xcite proved that the bifurcation locus in the space of real cubic polynomials is not locally connected either .
this makes the problem of defining a combinatorial model of @xmath134 very delicate .
there is no hope that a combinatorial model would lead to a precise topological model .
schleicher @xcite constructed a geodesic lamination modeling the space of _ unicritical _ polynomials , that is , polynomials with a unique multiple critical point .
we have heard of an unpublished old work of d. ahmadi and m. rees , in which cubic geodesic laminations were studied , however we have not seen it .
the present paper is based on the results obtained in @xcite .
these results are applicable to invariant laminations of any degree .
the paper is organized as follows . in section [ s :
basicdef ] , we discuss basic properties of geodesic laminations and laminational equivalence relations . in section [
s : long ] , we recall the results of @xcite adapting them to the cubic case .
finally , section [ s : cudebi ] is dedicated to the proof of the main result .
by a _ chord _ we mean a closed segment connecting two points of the unit circle .
if these two points coincide , then the chord is said to be _
[ d : geolam ] a _ geodesic lamination _ is a collection @xmath135 of chords called _ leaves _ that are pairwise disjoint in @xmath22 ; it includes all degenerate chords , and must be such that @xmath136 is closed .
_ gaps _ of @xmath135 are the closures of the components of @xmath137 .
we now introduce the notion of a ( sibling ) _
@xmath37-invariant geodesic lamination_. this is a slight modification of an invariant geodesic lamination introduced by thurston @xcite .
when @xmath0 is fixed we will often write `` invariant '' instead of `` @xmath37-invariant '' .
[ d : sibli ] a geodesic lamination @xmath135 is ( sibling ) _
@xmath37-invariant _ provided that : 1 . for each @xmath138
, we have @xmath139 , 2 .
[ 2]for each @xmath138 there exists @xmath140 so that @xmath141 .
3 . [ 3 ] for each @xmath138 such that @xmath142 is a non - degenerate leaf , there exist @xmath0 * pairwise disjoint * leaves @xmath143 , @xmath144 , @xmath145 in @xmath135 such that @xmath146 and @xmath147 for all @xmath148 , @xmath144 , @xmath0 .
observe that , since leaves are chords , and chords are closed segments , pairwise disjoint leaves in part ( 3 ) of the above definition can not intersect even on the unit circle ( that is , they can not have common endpoints ) .
call the leaf @xmath149 in ( [ 2 ] ) a _ pullback _ of @xmath150 .
a _ sibling of @xmath150 _ is defined as a leaf @xmath151 with @xmath152 .
thus , part ( 3 ) of the definition implies that any leaf with non - degenerate image has @xmath153 pairwise disjoint siblings .
moreover , these siblings can be chosen to be disjoint from the leaf .
definition [ d : sibli ] implies thurston s but is slightly more restrictive @xcite . from now on , by _
@xmath154-@xmath99invariant geodesic laminations _ , we always mean _ sibling @xmath37-invariant geodesic laminations_. moreover , for brevity we often talk about laminations meaning _ sibling @xmath37-invariant geodesic _ laminations .
[ d : linchor ] two * distinct * chords of @xmath22 are _ linked _ if they intersect in @xmath22 .
we will also sometimes say that these chords _ cross each other_. otherwise two chords are said to be _
unlinked_. a gap @xmath155 is said to be _
infinite @xmath98finite , uncountable@xmath99 _ if @xmath156 is infinite ( finite , uncountable ) .
uncountable gaps are also called _
_ gaps . for a closed convex set @xmath157 , straight segments in the boundary @xmath158 of @xmath159 are called _ edges _ of @xmath159 .
[ d : cristuff ] a _ critical chord @xmath98leaf@xmath99 _ @xmath45 of @xmath135 is a chord ( leaf ) of @xmath135 such that @xmath160 .
a gap is _ all - critical _ if all its edges are critical .
an all - critical gap or a critical leaf ( of @xmath135 ) is called an _ all - critical set _ ( of @xmath135 ) .
a gap @xmath155 of @xmath135 is said to be _ critical _ if it is an all - critical gap or there is a critical chord contained in the interior of @xmath155 except for its endpoints .
critical set _ of @xmath135 is by definition a critical leaf or a critical gap .
we also define a _ critical object _ of @xmath135 as a maximal by inclusion critical set . given a compact metric space @xmath161 , the space of all its compact subsets with the hausdorff metric is denoted by @xmath162 .
note that @xmath163 is compact for every geodesic lamination @xmath135 .
hence it can be viewed as a point in the space @xmath164 .
the family of sets @xmath163 of all invariant geodesic laminations @xmath135 is compact in @xmath164 .
however the set @xmath163 does not always determine the geodesic lamination @xmath135 .
indeed , any geodesic lamination @xmath135 without gaps has @xmath165 .
on the other hand , already in the cubic case there are two distinct invariant geodesic laminations @xmath166 and @xmath167 without gaps . here
@xmath166 contains all vertical chords and @xmath167 contains all horizontal chords .
observe that the corresponding topological julia sets are arcs with induced topological polynomials being non - conjugate `` saw - tooth '' maps .
however , @xmath135 itself is a point of @xmath168 which determines all the leaves of @xmath135 .
for this reason we consider the family of all invariant geodesic laminations as a subspace of @xmath168 with hausdorff metric ; the notion of convergence of invariant geodesic laminations is understood accordingly .
[ t : sibliclos ] the family of all invariant geodesic laminations @xmath135 is compact in @xmath168 . in other words ,
if invariant geodesic laminations @xmath169 converge to a collection of chords @xmath135 in @xmath168 ( that is , each leaf of @xmath135 is the limit of a sequence of leaves from @xmath169 , and each converging sequence of leaves of @xmath169 converges to a leaf of @xmath135 ) , then @xmath135 is an invariant geodesic lamination itself .
geodesic laminations naturally appear in the context of laminational equivalence relations . [
d : lam ] an equivalence relation @xmath39 on the unit circle @xmath15 is said to be _ laminational _ if either @xmath15 is one equivalence class ( such laminational equivalence relations are called _ degenerate _ ) , or the following holds : ( e1 ) the graph of @xmath39 is a closed subset of @xmath170 ; ( e2 ) the convex hulls of distinct equivalence classes are disjoint ; ( e3 ) each equivalence class of @xmath39 is finite .
[ d : si - inv - lam ] a laminational equivalence relation @xmath39 is called ( @xmath37-)_invariant _ if : ( d1 ) @xmath39 is _ forward invariant _ : for a @xmath39-class @xmath42 ,
the set @xmath171 is a @xmath39-class ; ( d2 ) for any @xmath39-equivalence class @xmath42 , the map @xmath172 extends to @xmath15 as an orientation preserving covering map such that @xmath42 is the full preimage of @xmath171 under this covering map .
for an invariant laminational equivalence relation @xmath39 , consider the _
topological julia set _ @xmath173 and the _ topological polynomial _
@xmath174 induced by @xmath37 .
the quotient map @xmath175 semiconjugates @xmath37 with @xmath176 .
a laminational equivalence relation @xmath39 admits a * canonical extension over @xmath13 * whose non - trivial classes are convex hulls of classes of @xmath39 . by moore
s theorem , the quotient space @xmath177 is homeomorphic to @xmath13 .
we will still denote the extended quotient map from @xmath13 to @xmath177 by @xmath178 ; the corresponding point - preimages ( _ fibers _ ) are the convex hulls of @xmath39-classes . with any fixed identification between @xmath179 and @xmath13
, one can extend @xmath180 to a branched covering map @xmath181 of degree @xmath0 called a
_ topological polynomial _ too .
the complement @xmath182 of the unique unbounded component @xmath183 of @xmath184 is called the _ filled topological julia set_. define the _ canonical geodesic lamination @xmath44 generated by @xmath39 _ as the collection of edges of convex hulls of all @xmath39-classes and all points of @xmath15 .
[ t : qsib ] geodesic laminations @xmath44 generated by @xmath37-invariant laminational equivalence relations are invariant .
we now consider dendritic laminations and corresponding topological polynomials .
[ d : dendri - all ] an invariant geodesic lamination @xmath44 is called _ dendritic _ if all its gaps are finite .
then the corresponding topological julia set @xmath185 is a dendrite .
the laminational equivalence relation @xmath39 and the topological polynomial @xmath180 are said to be _
dendritic _ too .
recall that , by @xcite , with every dendritic polynomial @xmath3 one can associate a dendritic topological polynomial @xmath36 so that @xmath31 is monotonically semi - conjugate to @xmath186 .
by @xcite , for every dendritic topological polynomial @xmath187 , there exists a polynomial @xmath3 with @xmath188 .
below , we list some well - known properties of dendritic geodesic laminations .
[ perfect ] let @xmath135 be a geodesic lamination considered as a subset of @xmath164 .
then the maximal perfect subset @xmath189 of @xmath135 is called the _ perfect part _ of @xmath135 .
a geodesic lamination @xmath135 is called _ perfect _ if @xmath190 .
equivalently , this means that all leaves of @xmath135 are non - isolated in the hausdorff metric .
observe that @xmath189 must contain @xmath15 .
the following lemma is well - known .
[ l : perfectd ] dendritic geodesic laminations @xmath135 are perfect
. we will need corollary 6.6 of @xcite , which reads : [ c : cridisj ] let @xmath135 be a perfect invariant geodesic lamination
. then the critical objects of @xmath135 are pairwise disjoint and are either all - critical sets , or critical sets whose boundaries map exactly @xmath191-to-@xmath192 , @xmath193 , onto their images . by , applies to dendritic geodesic laminations .
moreover , by properties of dendritic geodesic laminations , all their critical objects are finite . finally , let us state a property that follows from .
if @xmath135 is a dendritic geodesic lamination , and @xmath194 is an invariant geodesic lamination , then it follows that @xmath195 can be obtained from @xmath135 by inserting leaves in some grand orbits of gaps of @xmath135 .
moreover , as long as this insertion is done in a dynamically consistent fashion ( added leaves do not cross and form a fully invariant set ) , the new collection of leaves @xmath195 is an invariant geodesic lamination . on the other hand ,
the fact that @xmath195 is closed follows from the fact that if gaps of @xmath135 converge , then they must converge to a leaf of @xmath135 .
now we will review results of @xcite that are essential for this paper .
let us emphasize that results of @xcite hold for any degree .
however , we will adapt them here to degree three , omitting the general formulations .
consider a quadratic lamination @xmath135 with a critical quadrilateral @xmath196 .
thurston @xcite associates to @xmath135 its minor @xmath197 .
then @xmath198 is the full @xmath199-preimage of @xmath200 .
thurston proves that different minors obtained in this way never cross in @xmath22 .
observe that two minors cross if and only if the vertices of the corresponding critical quadrilaterals strictly alternate in @xmath15 .
thurston s result can be translated as follows in terms of critical quadrilaterals . if two quadratic laminations generated by laminational equivalences have critical quadrilaterals whose vertices strictly alternate , then the two laminations are the same .
this motivates definition [ d : strolin ] .
[ d : strolin ] let @xmath201 and @xmath202 be two quadrilaterals .
say that @xmath201 and @xmath202 are _ strongly linked _ if the vertices of @xmath201 and @xmath202 can be numbered so that @xmath203 where @xmath204 , @xmath205 , are vertices of @xmath201 and @xmath206 , @xmath205 are vertices of @xmath202 . since we want to study critical quadrilaterals in the degree three case , we now give a general definition .
[ d : qll ] a _ ( generalized ) critical quadrilateral _
@xmath196 is a circularly ordered quadruple @xmath207 $ ] of points @xmath208 in @xmath15 , where @xmath209 and @xmath210 are critical chords called _ spikes _ ; critical quadrilaterals @xmath207 $ ] , @xmath211 $ ] , @xmath212 $ ] and @xmath213 $ ] are viewed as equal .
we will often say `` critical quadrilateral '' when talking about the convex hull of a critical quadrilateral . clearly ,
if all vertices of a critical quadrilateral are distinct , or if its convex hull is a critical leaf , then the quadrilateral is uniquely defined by its convex hull . however ,
if the convex hull is a triangle , this is no longer true .
for example , let @xmath214 be an all - critical triangle .
then @xmath215 $ ] is a critical quadrilateral , but so are @xmath216 $ ] and @xmath217 $ ] .
if all vertices of a critical quadrilateral @xmath196 are pairwise distinct , then we call @xmath196 _ non - degenerate_. otherwise @xmath196 is called _
degenerate_. vertices @xmath218 and @xmath219 ( @xmath220 and @xmath221 ) are called _
opposite_. considering invariant geodesic laminations , all of whose critical sets are critical quadrilaterals , is not very restrictive
: we can `` tune '' a given invariant geodesic lamination by inserting new leaves into its critical sets in order to construct a new invariant geodesic lamination with all critical objects being critical quadrilaterals .
[ l : qls ] the family of all critical quadrilaterals is closed .
the family of all critical quadrilaterals that are critical sets of invariant geodesic laminations is closed too . being strongly linked is a closed condition on two quadrilaterals : if two sequences of critical quadrilaterals @xmath222 , @xmath223 are such that @xmath222 and @xmath223 are strongly linked and @xmath224 , @xmath225 , then @xmath201 and @xmath202 are strongly linked critical quadrilaterals . in @xcite ,
linked invariant geodesic laminations with quadratically critical portraits are defined for any degree @xmath0 .
below , we adapt this in the case of cubic laminations .
let @xmath135 be a cubic geodesic lamination ; in fact , in what follows when talking about laminations we always mean _ cubic _ laminations ( unless explicitly stated otherwise ) .
consider critical quadrilaterals @xmath226 , @xmath227 that are leaves or gaps of @xmath135 .
the pair @xmath228 is called a _ quadratically critical portrait _ of @xmath135 if @xmath226 and @xmath227 are distinct .
the triple @xmath229 is then called a _
cubic lamination with quadratically critical portrait_. sometimes , when the quadratically critical portrait is fixed , we write @xmath135 instead of @xmath229 .
observe that not all cubic geodesic laminations admit quadratically critical portraits .
for example , if @xmath135 has a unique critical object that is not all - critical , then @xmath135 has no quadratically critical portrait .
if @xmath135 has two disjoint critical objects , then @xmath226 , @xmath227 must coincide with these objects .
in particular , a cubic geodesic lamination with two disjoint critical objects admits a quadratically critical portrait if and only if both critical objects are ( possibly degenerate ) critical quadrilaterals .
assume that @xmath135 has an all - critical triangle @xmath230 .
then possible quadratically critical portraits of @xmath135 are : 1 .
pairs of distinct edges of @xmath230 ; and 2 . pairs
consisting of @xmath230 and an edge of it .
now we define linked laminations with quadratically critical portraits .
[ d : qclink1 ] let @xmath231 and @xmath232 be cubic invariant geodesic laminations with quadratically critical portraits .
these two laminations are said to be _ linked or essentially equal _
if one of the following holds : 1 . for every @xmath233 , the quadrilaterals @xmath234 and @xmath235 are strongly linked .
if @xmath234 and @xmath235 share a spike for every @xmath233 , then @xmath236 and @xmath237 are said to be essentially equal .
the laminations @xmath236 and @xmath237 share the same all - critical triangle .
then @xmath236 and @xmath237 are also said to be essentially equal . if ( 1 ) or ( 2 ) holds but @xmath236 , @xmath237 are not essentially equal , then @xmath236 , @xmath237 are said to be _ linked_. we can also talk about linked or essentially equal quadratically critical portraits without referring to laminations containing them .
critically marked polynomials , topological polynomials , and laminational equivalence relations were defined in the introduction ; recall that there they are all assumed to be dendritic .
let us now define critically marked cubic _
geodesic _ laminations .
suppose that @xmath135 is a cubic geodesic lamination and an ordered pair of critical sets ( gaps or leaves ) @xmath238 of @xmath135 is chosen so that on the boundary of each component @xmath239 of @xmath240 the map @xmath113 is one - to - one ( except for the endpoints of possibly existing critical edges of such components ) .
then we can consider @xmath241 as a _
( critically ) marked _ lamination even though @xmath135 is not necessarily dendritic .
for brevity we often talk about _ marked _
( topological ) polynomials and laminations rather than _ critically marked _ ones .
let @xmath242 be a marked cubic geodesic lamination .
then @xmath243 is called a _ critical pattern _ of @xmath135 ; when talking about _ critical patterns _ we mean critical patterns of some marked lamination @xmath135 and allow for @xmath135 to be unspecified .
let @xmath135 be a dendritic lamination . then if @xmath109 are its critical sets , the only two possible critical patterns that can be associated with @xmath135 are @xmath244 or @xmath245 .
now , if @xmath135 has a unique critical set @xmath161 which is not an all - critical triangle then the only possible critical pattern of @xmath135 is @xmath246 .
however if @xmath135 has a unique critical set @xmath230 which is an all - critical triangle then there are more possibilities for a critical pattern of @xmath135 .
namely , by definition a critical pattern of @xmath135 can be either @xmath247 , of @xmath230 and an edge of @xmath230 , or an edge of @xmath230 and @xmath230 , or an ordered pair of two edges of @xmath230 .
we defined linked or essentially equal cubic laminations with quadratically critical portraits .
let us extend this notion to marked laminations and their critical patterns .
a _ collapsing quadrilateral _ is a critical quadrilateral that maps to a non - degenerate leaf .
[ d : link3 ] marked laminations @xmath248 and @xmath249 , and their critical patterns @xmath250 and @xmath251 , are said to be _ linked _ ( _ essentially equal _ ) if there are quadratically critical portraits @xmath252 @xmath253 and @xmath254 such that @xmath255 for which if @xmath256 is a collapsing quadrilateral , then it shares a pair of opposite edges with @xmath257 .
let @xmath196 be one of @xmath256 , and @xmath106 be the corresponding @xmath257 .
then the image of @xmath196 is an edge of @xmath258 .
hence forward images of quadrilaterals @xmath256 will never form linked leaves . pulling back the sets @xmath256
, we construct the `` tuned '' laminations @xmath259 and @xmath260 with quadratically critical portraits @xmath261 @xmath253 and @xmath254 .
the existence of `` tuned '' laminations with quadratically critical portraits follows from our definitions and thurston s pullback construction .
lemma [ l : actri ] follows immediately from definitions .
[ l : actri ] if two limit laminations contain the same all - critical triangle , then they are essentially equal .
the following is a special case of one the central results of @xcite ( in @xcite more general results are obtained for all degrees ) .
[ t : noesli ] let @xmath248 and @xmath249 be marked laminations .
suppose that @xmath236 is perfect ( e.g. , by lemma [ l : perfectd ] @xmath236 may be dendritic ) . if they are linked or essentially equal then @xmath262 and @xmath263 for @xmath233 .
in particular , if both laminations are perfect , then @xmath264 .
in the rest of the paper , we define a visual parameterization of the family of all marked cubic dendritic geodesic laminations .
let @xmath265 be a sequence of marked laminations with critical patterns @xmath266 where @xmath267 are finite .
assume that the sequence @xmath169 converges to an invariant lamination @xmath268 ( see theorem [ t : sibliclos ] ) ; then the critical sets @xmath269 , @xmath270 converge to gaps ( or leaves ) @xmath271 of @xmath268 .
we say that the sequence @xmath265 _ converges _ to @xmath272 .
every vertex of @xmath274 has a sibling vertex in @xmath274 .
it follows that @xmath274 is critical .
if @xmath274 is periodic of period , say , @xmath275 , then , since it is critical , it is an infinite gap . then the fact that @xmath276 implies that any gap @xmath269 sufficiently close to @xmath274 will have its @xmath277-image also close to @xmath274 , and therefore coinciding with @xmath269 .
thus , @xmath269 is @xmath113-periodic , which is impossible because @xmath269 is finite and critical .
similarly , @xmath278 is critical and non - periodic .
let us show that @xmath273 is a marked lamination . to this end
we need to show that on the boundary of each component @xmath239 of @xmath279 the map @xmath113 is one - to - one ( except for the endpoints of possibly existing critical edges of such components ) .
however this follows from definitions and the fact that the same claim holds for all @xmath280 . any marked lamination similar to @xmath281 from lemma [
l : decrilim ] will be called a _
limit marked lamination_. in particular , a marked dendritic lamination is a limit marked lamination ( consider a constant sequence ) .
recall that for a compact metric space @xmath161 , the space of all its compact subsets with the hausdorff metric is denoted by @xmath162 . as was explained in the introduction
, a marked cubic dendritic polynomial always defines a marked cubic lamination .
take a marked dendritic polynomial @xmath282 and let @xmath242 be the corresponding marked lamination .
define the map @xmath283 by setting @xmath284 .
consider a sequence of marked dendritic cubic geodesic laminations @xmath285 .
if @xmath169 converge then , by theorem [ t : sibliclos ] , the limit @xmath268 is itself a cubic geodesic lamination , and by the above the critical patterns @xmath286 converge to the critical pattern @xmath287 of @xmath268 .
we are interested in the case when @xmath288 is in a sense compatible with a dendritic lamination .
lemma [ l : uppers ] says that if critical patterns of dendritic cubic geodesic laminations converge * into * a critical pattern of a dendritic cubic geodesic lamination @xmath135 , then the limit lamination contains @xmath135 .
[ c : crista ] suppose that a sequence @xmath292 @xmath293 of marked cubic dendritic polynomials converges to a marked cubic dendritic polynomial @xmath294 .
consider corresponding marked laminational equivalence relations @xmath295 and @xmath296 .
if @xmath297 converges to @xmath298 @xmath299 @xmath300 , then we have @xmath301 in particular , the map @xmath302 is upper semi - continuous . by corollary [ c : crista ] ,
critical objects of dendritic invariant geodesic laminations @xmath303 associated with polynomials @xmath124 can not explode under perturbation of @xmath3 ( they may implode though ) .
[ d : admicri ] let @xmath106 be a leaf or a gap of a cubic invariant geodesic lamination @xmath135 .
assume that either @xmath106 is the only critical object of @xmath135 , or there is exactly one hole of @xmath106 of length @xmath306 .
if @xmath106 is the only critical object of @xmath135 , we set @xmath307 .
otherwise let @xmath159 be the unique hole of @xmath106 of length @xmath308 , let @xmath201 be the set of all points in @xmath159 with the images in @xmath258 , and set @xmath309 .
the set @xmath310 is called the _ co - critical set _ of @xmath106 .
sets @xmath312 ( and hence mixed tags ) are well - defined .
the mixed tag @xmath313 of a marked lamination is the product of two sets , each of which is a point , a leaf , or a gap .
one can think of @xmath314 as a higher dimensional analog of a gap / leaf of a geodesic lamination .
we show that the union of tags of marked dendritic laminations is a ( non - closed ) `` geodesic lamination '' in @xmath315 .
the main idea is to relate the non - disjointness of mixed tags of marked dendritic laminations and their limits with the fact that they have `` tunings '' that are linked or essentially equal .
clearly , a unicritical marked lamination has a unique critical object .
however a lamination @xmath135 with unique critical object may have a bicritical critical pattern . by definition
this is only possible if @xmath135 has an all - critical gap @xmath230 and the critical pattern is either two edges of @xmath230 or @xmath230 and an edge of @xmath230 .
[ l : interlink ] suppose that two marked laminations have non - disjoint mixed tags .
suppose also that at least one of the two laminations is dendritic . then either the two marked laminations are linked or essentially equal , or the geodesic laminations are equal and share an all - critical triangle .
the proof of lemma [ l : interlink ] is mostly non - dynamic and involves checking various cases .
we split the proof into propositions .
observe that mixed tags are determined by critical patterns ; we do not need laminations to define mixed tags . in propositions [ p :
mnondeg ] - [ p : quadfup ] we assume that the critical patterns @xmath317 and @xmath318 are bicritical and have non - disjoint mixed tags . by the assumption ,
some distinct edges of the sets @xmath319 and @xmath320 cross .
denote these linked edges by @xmath321 and @xmath322 , see fig .
[ f : mnondeg ] .
we may choose the orientation so that @xmath323 , @xmath324 are in the holes of @xmath319 , @xmath320 disjoint from @xmath325 , @xmath326 respectively , and so that @xmath327 .
we claim that @xmath323 is of length at most @xmath328 .
indeed , if @xmath323 had length greater than @xmath328 , then there would exist a sibling @xmath150 of @xmath321 with endpoints in @xmath323 . evidently , @xmath150 would be an edge of @xmath329 , contradicting the choice of @xmath323 .
thus , @xmath323 is of length at most @xmath328 and the restriction @xmath330 is one - to - one .
similarly , @xmath324 is of length at most @xmath328 and the restriction @xmath331 is one - to - one .
let us show now that @xmath332 $ ] .
if @xmath333 is of degree three , this follows immediately .
otherwise , let @xmath334 and @xmath335 .
. moreover , since @xmath329 is critical , vertices of @xmath329 partition the arc @xmath337 into open arcs on each of which the map is one - to - one .
hence @xmath338 must have vertices in @xmath339\cup[b_1 , a_1']$ ] .
since each of these intervals maps onto @xmath340 $ ] one - to - one , our claim follows .
we claim that @xmath341 . indeed , otherwise @xmath342\subset [ a_2,b_2)$ ] , which implies that @xmath340\subset [ { \sigma}_3(a_2 ) , { \sigma}_3(b_2))$ ] .
on the other hand , by the above we have @xmath343 $ ] and @xmath344 $ ] . since @xmath345 , we have in fact @xmath346 , a contradiction .
thus , the points @xmath204 and @xmath206 for @xmath347 belong to an arc of length at most @xmath328 . we claim that then @xmath348 and @xmath349 are strongly linked collapsing quadrilaterals
indeed , we have that @xmath350 .
it follows then that @xmath351 and therefore that , indeed , @xmath352 and @xmath353 are strongly linked collapsing quadrilaterals .
moreover , since @xmath321 and @xmath322 are edges of @xmath354 and @xmath320 it follows that the quadrilateral @xmath352 shares two edges with the set @xmath329 , and the quadrilateral @xmath355 shares two edges with the set @xmath326 .
note that all vertices of @xmath338 and @xmath356 are in @xmath357\cup [ b'_2,a_1]$ ] , where @xmath358 and @xmath359 .
the restriction of @xmath113 to each of the arcs @xmath357 $ ] , @xmath360 $ ] is injective .
therefore , a pair of linked edges ( or a pair of coinciding vertices ) of @xmath361 and @xmath362 gives rise to a pair of linked quadrilaterals @xmath363 and @xmath364 in @xmath338 and @xmath356 , respectively , so that if these quadrilaterals share edges with containing them critical sets .
we will use the same notation as in the proof of proposition [ p : mnondeg ] .
if @xmath319 and @xmath320 have distinct edges that cross in @xmath22 , then proposition [ p : mnondeg ] applies .
assume now that @xmath319 and @xmath320 share a vertex @xmath46 .
clearly , there is a unique critical chord @xmath150 such that @xmath366
. then @xmath367 , and we may set @xmath368 . both sets @xmath270 have vertices in the closed arc @xmath201 of length @xmath369 bounded by the endpoints of @xmath150 .
by our assumption , @xmath370 .
if the sets @xmath371 have a pair of linked edges or share a vertex @xmath372 , then these edges or @xmath95 can by pulled back to @xmath373 as a pair of linked critical quadrilaterals .
assume now that @xmath374 .
clearly , @xmath375 . set @xmath376 .
we claim that @xmath230 is a gap of @xmath236 . indeed
, the set @xmath338 contains at least two vertices of @xmath230 and is non - disjoint from @xmath325 . since @xmath236 is dendritic , @xmath236 has a unique critical object @xmath239 . if @xmath377 , then by definition the critical pattern of @xmath236 is @xmath378 , a contradiction with the assumption that @xmath236 is bicritical .
thus , @xmath230 is a gap of @xmath236 .
we claim that @xmath230 is a gap @xmath237 .
we prove first that there is an edge @xmath379 of @xmath230 such that one of the sets @xmath380 contains @xmath150 while the other one contains @xmath149 .
this is obvious if @xmath356 contains an edge @xmath379 of @xmath230 .
otherwise @xmath381 .
then @xmath150 must be an edge of @xmath356 because otherwise the sets @xmath326 and @xmath356 will have either non - disjoint interiors , or one of them is contained in the interior of the other one , a contradiction .
similarly , @xmath150 is an edge of @xmath326 .
it follows that one of the sets @xmath382 is @xmath150 while the other one is a critical gap @xmath383 with @xmath150 as an edge . by the above , @xmath150 and @xmath149
are either leaves of @xmath237 or are contained in gaps of @xmath237 .
moreover endpoints of @xmath150 and @xmath149 are not periodic since @xmath230 is a gap of a dendritic lamination @xmath236 . hence @xmath150 and @xmath149 can be pulled back in a unique way and its pullbacks either will be contained in gaps of @xmath237 or will be leaves of @xmath237 .
this yields a new lamination @xmath384 and a marked lamination @xmath385 ) .
consider also the marked lamination @xmath386 .
since these two marked laminations are essentially equal , theorem [ t : noesli ] implies that @xmath387 . hence @xmath230 is a gap of @xmath388 and , moreover , leaves shared by @xmath236 and @xmath388 approximate all edges of @xmath230 from outside of @xmath230 .
it follows that @xmath230 is a subset of a gap @xmath155 of @xmath237 .
let us show that @xmath389 .
by lemma [ l : decrilim ] , @xmath155 is not periodic .
hence pullbacks of @xmath150 and @xmath149 do not re - enter @xmath155 , and so an edge of @xmath230 contained in the interior of @xmath155 ( except for the endpoints ) remains isolated in both @xmath237 and @xmath388 .
however in the previous paragraph we concluded that it is not isolated in @xmath388 , a contradiction .
we conclude that @xmath230 is a gap of @xmath237 .
let us show that @xmath365 .
we can adjust the critical pattern of @xmath237 so that it coincides with the critical pattern of @xmath236 . by theorem [ t : noesli ]
, we then have @xmath390 .
moreover , no leaves of @xmath237 are contained in the unique critical set @xmath230 of @xmath236 .
by @xcite , any periodic gap of @xmath236 has a single cycle of edges .
we conclude that no leaves of @xmath237 are contained in periodic or preperiodic gaps of @xmath236 .
finally , by @xcite there are no wandering gaps of @xmath236 .
this implies that @xmath391 , as claimed .
[ l : uniclink ] suppose that @xmath392 and @xmath393 are marked unicritical laminations with non - disjoint mixed tags . then @xmath248 and @xmath394 are linked or essentially equal where @xmath257 either equals @xmath395 or is a critical chord contained in @xmath395 .
suppose that @xmath236 has an all - critical triangle @xmath230 ( and so @xmath396 ) . since the mixed tags intersect , then @xmath397 and hence @xmath398 .
choosing two edges of @xmath230 as a quadratically critical portrait in @xmath399 and in @xmath400 , we see that by definition @xmath248 and @xmath394 are essentially equal .
suppose that neither invariant geodesic lamination has an all - critical triangle .
if @xmath401 contains a point @xmath402 , then the entire all - critical triangle @xmath403 is contained in @xmath404 ; we can choose the same two edges of @xmath230 as a quadratically critical portrait for both laminations .
otherwise , we may assume that an edge @xmath143 of @xmath405 crosses an edge @xmath406 of @xmath407 .
this implies that the hexagons @xmath408 and @xmath409 have alternating vertices and proves the lemma in this case too
. denote laminations in question by @xmath236 and @xmath237 .
if both laminations are bicritical , then the result follows from proposition [ p : quadfup ] .
if both laminations are unicritical , then the result follows from lemma [ l : uniclink ] .
it remains to consider the case where the first critical pattern @xmath410 is unicritical , and the second one @xmath411 is bicritical .
either an edge of @xmath405 crosses an edge of @xmath362 or a vertex of @xmath405 lies in @xmath362 . in either case , there are two sibling edges or sibling vertices of @xmath356 that are linked or coincide with edges or vertices of @xmath399 .
taking convex hulls of these pairs of ( possibly degenerate ) leaves , we obtain @xmath412 and @xmath413 . by construction , these are strongly linked quadrilaterals . similarly , there is a ( possibly degenerate ) leaf in @xmath320 that is linked or equal to a leaf in @xmath399 .
it follows that the two siblings of this leaf in @xmath326 are linked or equal to some leaves in @xmath399 . as above
, this leads to strongly linked quadrilaterals @xmath414 and @xmath415 .
it is easy to see that @xmath416 and @xmath254 are linked or essentially equal quadratically critical portraits . 1 .
@xmath365 has an all - critical triangle @xmath230 , it is not true that @xmath325 and @xmath326 are distinct edges of @xmath230 , and either @xmath419 , or @xmath420 . ; 2 .
there is no all - critical triangle in @xmath262 , and @xmath263 for @xmath233 ( in particular , if @xmath237 is dendritic then @xmath365 ) .
if the mixed tags of @xmath248 and @xmath421 are non - disjoint , then , by lemma [ l : interlink ] , either @xmath365 share an all - critical triangle @xmath230 , or these marked laminations are linked or essentially equal . in the first case
consider several possibilities for the critical patterns .
one can immediately see that the only way the mixed tags are disjoint is when @xmath325 and @xmath326 are distinct edges of @xmath230 ; since the mixed tags are known to be non - disjoint we see that this corresponds to case ( 1 ) from the theorem . in the second case
the fact that our marked laminations are linked or essentially equal implies , by theorem [ t : noesli ] , that case ( 2 ) of the theorem holds .
the opposite direction of the theorem follows from definitions .
[ d : usc ] a collection @xmath422 of compact and disjoint subsets of a metric space @xmath161 is _ upper semi - continuous @xmath98usc@xmath99 _ if , for every @xmath423 and every open set @xmath424 , there exists an open set @xmath425 containing @xmath423 so that , for each @xmath426 , if @xmath427 , then @xmath428 . a decomposition of a metric space
is said to be _ upper semi - continuous @xmath98usc@xmath99 _ if the corresponding collection of sets is upper semi - continuous . before applying theorem [ t : dav ] to our tags
, we draw a distinction between two notions .
if @xmath39 is a dendritic laminational equivalence relation , and @xmath44 has critical objects @xmath431 then by a _ marked laminational equivalence relation _ we mean a triple @xmath432 or a triple @xmath433 ( if @xmath434 ) , and just @xmath435 ( if @xmath436 ) .
consider a marked geodesic lamination @xmath437 .
each marked laminational equivalence relation is associated with the corresponding marked geodesic lamination ( the first of the two critical sets in a marked laminational equivalence relation becomes @xmath399 and the second becomes @xmath400 ) .
however if @xmath39 has an all - critical triangle @xmath230 then there are more possibilities for @xmath437 than just @xmath438 .
e.g. , @xmath439 could be two distinct edges of @xmath230 .
still , mixed tags of laminational equivalence relations are mixed tags of the corresponding geodesic laminations and so our results obtained for geodesic laminations apply to them .
recall that the map @xmath118 was defined in definition [ d : siblita ] . to a marked laminational equivalence relation @xmath110 , or to its critical pattern @xmath244 , the map @xmath118 associates the corresponding _ mixed tag _ @xmath440 .
[ t : tagusc ] the family @xmath441 of mixed tags of cubic marked dendritic invariant laminational equivalence relations forms an upper semi - continuous decomposition of the union @xmath129 of all these tags .
if @xmath442 and @xmath443 are marked dendritic laminational equivalence relation , and @xmath444 and @xmath445 are non - disjoint , then , by theorem [ t : dendrilink ] applied to the marked geodesic laminations @xmath446
@xmath447 @xmath448 and @xmath449 , we have that the corresponding marked laminational equivalence relations are equal , i.e. @xmath450 . hence the family @xmath451 @xmath452 forms a decomposition of @xmath129 .
suppose next that @xmath453 is a sequence of marked dendritic laminational equivalence relations with @xmath454 .
assume that there is a limit point of the sequence of their tags @xmath455 that belongs to the tag of a marked dendritic laminational equivalence @xmath456 where @xmath457 .
by @xcite and lemma [ l : decrilim ] we may assume that the sequence @xmath458 converges to a marked lamination @xmath273 with critical pattern @xmath459 . by the assumption , @xmath460 . by theorem [ t :
dendrilink ] , @xmath461 and @xmath462 for @xmath233 .
hence @xmath463 .
denote the quotient space of @xmath129 , obtained by collapsing every element of @xmath126 to a point , by @xmath464 ( elements of @xmath126 are mixed tags of critical patterns of marked dendritic laminational equivalence relations ) .
let @xmath465 be the quotient map . by theorem [ t : dav ]
, the topological space @xmath464 is separable and metric .
we show that @xmath464 can be viewed as a combinatorial model for @xmath103 .
recall that the map @xmath283 was defined right before lemma [ l : uppers ] . by definition and corollary [ c : crista ] ,
the map @xmath302 is upper semi - continuous and surjective . also , @xmath118 is continuous in the hausdorff topology and preserves inclusions .
finally , @xmath467 is continuous by definition .
thus , @xmath468 is a continuous surjective map as desired .
a. douady , _ descriptions of compact sets in @xmath13 _ , topological methods in modern mathematics ( stony brook , ny , 1991 ) , ed . by l. goldberg and a. phillips ,
publish or perish , houston , tx ( 1993 ) , 429465 .
d. schleicher , _ on fibers and local connectivity of mandelbrot and multibrot sets _ , fractal geometry and applications : a jubilee of benoit mandelbrot , ed . by m. lapidus and m. van frankenhuysen , proc .
pure math . * 72 * , part 1 , amer .
soc . , providence , ri ( 2004 ) , 477517 .
|
complex 1-variable polynomials with connected julia sets and only repelling periodic points are called _ dendritic_. by results of kiwi , any dendritic polynomial is semi - conjugate to a topological polynomial whose topological julia set is a dendrite .
we construct a continuous map of the space of all cubic dendritic polynomials onto a laminational model that is a quotient space of a subset of the closed bidisk .
this construction generalizes the `` pinched disk '' model of the mandelbrot set due to douady and thurston .
it can be viewed as a step towards constructing a model of the cubic connectedness locus .
|
the formation of a hot and dense medium of deconfined quarks and gluons ( qgp ) has been established in high energy heavy ion collisions at the relativistic heavy ion collider ( rhic ) facility at brookhaven national laboratory and the large hadron collider ( lhc ) facility at cern @xcite .
the transition from qgp to a hadron gas has been shown to be a cross over @xcite .
the focus of research in this field has now shifted towards two aspects , ( a ) characterizing the transport properties of qgp and ( b ) establishing the qcd phase structures at high baryonic chemical potential ( @xmath6 ) region of the qcd phase diagram .
a rigorous phenomenological analysis of the precision data from relativistic heavy - ion collisions and theoretical advances over 14 years has led to quantitative estimates for some of the transport properties of a strongly interacting de - confined state of quarks and gluons .
the estimated shear viscosity to entropy density ratio ( @xmath7 ) is found to reflect the inviscid liquid property of the qgp and has a value between ( 1 - 2)/4@xmath8 @xcite . that reflecting the stopping power or the opacity of qgp has been estimated by obtaining the square of the momentum transferred by the parton to the qgp per unity length ( @xmath9 ) and
is found to lie between 2 - 10 gev@xmath10/fm @xcite . on the other hand
a dedicated program called the beam energy scan ( bes ) to establish the phase diagram of qcd was launched at rhic in the year 2010 to unravel the qcd phase structure at large @xmath6 .
a range of @xmath6 from 20 mev to 400 mev of the phase diagram was covered by varying the @xmath11 from 200 to 7.7 gev .
the rich phase structure in the high @xmath6 region can be seen from the conjectured qcd phase diagram shown in fig .
[ phasediagram ] @xcite .
the two distinct features of the phase diagram at large @xmath6 are the first order phase boundary and the critical point ( cp ) . in this paper
we concentrate on the status of the experimental search for these two phase structures through the measurement of proton and antiproton production in heavy - ion collisions .
specifically we discuss the observable related to the azimuthal and multiplicity distributions for net - proton ( the difference in number of protons and anti - protons ) in au+au collisions at midrapidity for @xmath0 = 7.7 , 11.5 , 19.6 , 27 , 39 , 62.4 and 200 gev .
we find the results to be very intriguing , which must to be further quantified by having a high event statistics second phase of bes program in near future at rhic . in the next section
we discuss the two observables related to the search for the first order phase transition and the critical point . in section 3
we present the experimental results on the directed flow measurements of net - protons , an observable for first order phase transition . in section 4
we discuss the measurements related to the product of higher moments of net - proton multiplicity distribution , observable for critical point search .
finally in section 5 we summarize the findings .
the experimental results presented here are from the data recorded in the star detector at rhic in the year 2010 and 2011 .
the patterns of azimuthal anisotropy in particle production , often termed as flow , in heavy - ion collisions can be obtained by studying the fourier expansion of the azimuthal angle ( @xmath12 ) distribution of produced particles with respect to the reaction plane angle ( @xmath13 ) @xcite . where the reaction plane is defined as the plane subtended by the impact parameter direction and the beam direction .
the various ( order @xmath14 ) coefficients in this expansion are defined as : @xmath15\rangle.\ ] ] the angular brackets in the definition denote an average over many particles and events .
directed flow is quantified by the first coefficient ( @xmath16 ) . on the other hand
the elliptic flow is given by the second coefficient ( @xmath17 ) .
the @xmath16 , which is sensitive to early collision dynamics , is proposed as a signature of first order phase transition based on a hydrodynamic calculation @xcite .
these calculations whose equation of state incorporates a first - order phase transition from hadronic matter to qgp predict a non - monotonic variation of the slope of directed flow of baryons ( and net - baryons ) around midrapidity as a function of beam energy , also has a prominent minimum , and a double sign change in the @xmath18 slope , which is not seen in the same hydrodynamic model without a first - order phase transition .
the @xmath16 results discussed in this paper are for the most abundantly measured baryons , antiproton and proton , detected using their energy loss in star time projection chamber and by the time - of - flight information from the time of flight detector @xcite .
protons and antiprotons have a transverse momentum between 0.4 and 2.0 gev/@xmath19 and pseudorapidity ( @xmath20 ) between @xmath21 1 unit .
the first order event plane for @xmath11 @xmath22 62.4 gev is constructed using the information from the two beam beam counters at 3.3 @xmath22 @xmath23 @xmath22 5.0 .
that for @xmath11 = 62.4 and 200 gev uses the information from star zdc - smd detectors .
all results discussed here are corrected for event plane resolution and proton track reconstruction efficiency .
non - monotonic variations of observables related to the moments of the distributions of conserved quantities such as net - baryon , net - charge , and net - strangeness @xcite number with @xmath24 are believed to be good signatures of a cp .
the moments are related to the correlation length ( @xmath25 ) of the system @xcite .
finite size and time effects in heavy - ion collisions put constraints on the significance of the desired signals .
a theoretical calculation suggests a non - equilibrium @xmath25 @xmath26 2 - 3 fm for heavy - ion collisions @xcite .
hence , it is proposed to study the higher moments ( like skewness , @xmath27 = @xmath28 and kurtosis , @xmath29 = [ @xmath30 3 with @xmath31 = @xmath32 @xmath33 ) of distributions of conserved quantities due to a stronger dependence on @xmath25 @xcite .
further , products of the moments can be related to susceptibilities associated with the conserved numbers .
the product @xmath29@xmath34 of the net - baryon number distribution is related to the ratio of fourth order ( @xmath35 ) to second order ( @xmath36 ) baryon number susceptibilities @xcite .
the ratio @xmath35/@xmath36 is expected to deviate from unity near the cp .
it has different values for the hadronic and partonic phases @xcite .
the higher moments of the net - proton multiplicity ( @xmath37 = @xmath38 ) distributions from au+au collisions discussed in this paper are for protons and anti - protons detected at midrapidity ( @xmath39 @xmath22 0.5 ) in the range 0.4 @xmath22 @xmath40 @xmath22 0.8 gev/@xmath19 .
a good purity of the proton sample ( better than 98% ) for all beam energies is obtained in the momentum range studied .
all results presented are corrected for proton reconstruction efficiency .
) near mid - rapidity as a function of beam energy ( @xmath11 ) for intermediate - centrality ( 10 - 40% ) au+au collisions @xcite . panels ( a ) , ( b ) and ( c ) shows the star experiment measurement for antiprotons , protons , and net - protons , respectively , along with corresponding calculations from the urqmd model obtained with the same cuts and fit conditions . the systematic uncertainties on the measurements are shown as shaded bars .
the dashed curves are a smooth fit to guide the eye . ]
figure [ fig : v1 ] shows the slope of directed flow versus rapidity ( @xmath1 ) near midrapidity as a function of @xmath11 for antiprotons ( panel ( a ) ) , protons ( panel ( b ) ) and net - proton ( panel ( c ) ) @xcite .
the results are for 10 - 40% au+au collision centrality .
the antiproton @xmath1 has negative values and shows a monotonic increase with @xmath11 .
the proton @xmath1 has positive values for @xmath11 = 7.7 gev and negative values for rest of the energies studied .
the proton @xmath1 dependence on @xmath11 is non - monotonic with a minimum around @xmath11 = 11.5 and 19.6 gev .
the energy dependence of proton @xmath41 involves an interplay between the directed flow of protons associated with baryon number transported from the initial beam rapidity to the vicinity of mid - rapidity , and the directed flow of protons from particle - antiparticle pairs produced near mid - rapidity .
the importance of the pair production mechanism increases strongly with beam energy .
it is important to distinguish between the two mechanisms before further conclusions can be drawn .
the net - proton @xmath41 is expected to provide the contribution from protons associated with baryon number transport .
assuming antiproton directed flow as a proxy for the directed flow of pair produced protons , the proposed net - proton slope can be constructed from @xmath42_p = r(y ) [ v_1(y)]_{\bar{p } } + [ 1-r(y)]\ , [ v_1(y)]_{{\rm net\mbox{-}}p}\,,\ ] ] where @xmath43 is the observed rapidity dependence of the ratio of antiprotons to protons at each beam energy @xcite .
the net - proton slope is shown as a function of @xmath11 in fig .
[ fig : v1](c ) .
the data shows a non - monotonic dependence on @xmath11 with a minimum around @xmath11 = 11.5 and 19.6 gev .
the values of slope changes sign twice , goes from positive at 7.7 gev to negative at @xmath11 = 11.5 - 27 gev , then again becomes positive for @xmath11 @xmath44 39 gev .
the corresponding urqmd results @xcite , which does not include any first order phase transition effects , in all the three cases of slope of antiproton , proton and net - proton shows a monotonic variation with @xmath11 .
the slope values also does not agree with the measurements .
a possible interpretation of the changing sign of the @xmath18 slope is that it reflects a change in equation - of - state . at a given energy where the system undertakes a first order quark
hadron phase transition , one expect the formation of a mixed - phase for which the pressure gradient is small
. the softest pressure will naturally produces the observed minimum in @xmath18 slope parameter @xcite .
the alternate proposal is that at higher energies pair production is dominant at mid - rapidity and transported baryons have relatively small influence . as there is no preferred direction for pair produced hadrons ,
so the slope parameter becomes close to zero .
while at lower beam energies the observed baryons are strongly influenced by the transported baryons so they are aligned with them hence the slope parameter is positive .
the most intriguing is the dependence in the intermediate energies , a mean field model study shows that the energy dependent baryon potential plays an important role in this region @xcite . in the search for the signature of a first - order phase transition in the high @xmath6 region of the qcd phase diagram ,
the findings from @xmath1 from the rhic beam energy scan program are very compelling and strongly motivates further measurements .
to better understand the possible role and relevance of stopping in interpretation of existing data on net - proton directed flow , new higher event statistics measurements of the centrality dependence of @xmath18 at @xmath11 = 7.7 to 19.6 gev is needed .
and @xmath29@xmath34 from au+au and @xmath45+@xmath45 collisions at rhic @xcite .
crosses , open squares and filled circles are for the efficiency corrected results of @xmath45+@xmath45 , 70 - 80% , and 0 - 5% au+au collisions , respectively .
skellam distributions for corresponding collision centralities are shown in the top panel .
shaded hatched bands are the results from urqmd .
the hadron resonance gas model ( hrg ) values for @xmath29@xmath34 and @xmath46/skellam are unity .
the error bars are statistical and caps are systematic errors . for clarity , @xmath45+@xmath45 and 70 - 80% au+au results
are slightly displaced horizontally . ]
figure [ fig : cp ] shows the energy dependence of @xmath46 and @xmath29@xmath34 for @xmath38 for au+au collisions for two collision centralities ( 0 - 5% and 70 - 80% ) , corrected for @xmath45(@xmath47 ) reconstruction efficiency @xcite .
the skellam expectations for @xmath46 are calculated using the data as @xmath48 .
the @xmath46 values normalized to the corresponding skellam expectations are shown in the bottom panel of fig .
[ fig : cp ] .
the skellam expectations reflect a system of totally uncorrelated , statistically random particle production .
the central collision data shows deviation from skellam expectation with maximum deviation occurring for @xmath49 = 19.6 and 27 gev .
the corresponding results from @xmath45+@xmath45 collisions at @xmath24 = 62.4 and 200 gev are also shown and found to be similar to peripheral au+au collisions within the statistical errors . for @xmath49 = 19.6 and 27 gev
, differences are observed between the 0 - 5% central au+au collisions and the peripheral collisions .
the results are closer to unity for @xmath24 = 7.7 gev .
higher statistics data for @xmath24 @xmath22 19.6 gev will help in quantitatively understanding the suggestive non - monotonic energy dependence of @xmath29@xmath34 and @xmath46 .
the data also show deviations from the hadron resonance gas model @xcite which predict @xmath29@xmath34 and @xmath46/skellam to be unity . the effect of decay is less than 2% as per the hrg calculations in ref .
@xcite . to understand the effects of baryon number conservation @xcite and experimental acceptance , urqmd model calculations ( a transport model which does not include a cp ) @xcite for 0 - 5% au+au collisions
are shown in the middle and bottom panels of fig .
[ fig : cp ] .
the urqmd model shows a monotonic decrease with decreasing beam energy @xcite .
the centrality dependence of the @xmath29@xmath34 and @xmath46 from urqmd @xcite ( not shown in the figures ) closely follow the data at the lower beam energies of 7.7 and 11.5 gev .
their values are in general larger compared to data for the higher beam energies .
the observed @xmath24 dependence for central au+au collisions is not explained by urqmd model .
the current data provide the most relevant measurements over the widest range in @xmath6 ( 20 to 450 mev ) to date for the cp search , and for comparison with the baryon number susceptibilities computed from qcd to understand the various features of the qcd phase structure @xcite .
the deviations of @xmath50@xmath51 and @xmath29@xmath34 below skellam expectation are qualitatively consistent with a qcd based model which includes a cp @xcite .
however , conclusions on the existence of cp can be made only after a high event statistics measurement for @xmath11 @xmath22 27 gev in the second phase of the beam energy scan program and comparison to qcd calculations with cp behavior .
we have discussed two striking observations from the rhic beam energy scan program related to the first order quark
hadron phase transition and the critical point .
the measurements uses the produced protons and antiprotons in au+au collisions at midrapidity for @xmath11 = 7.7 , 11.5 , 19.6 , 27 , 39 , 62.4 and 200 gev .
the slope of the directed flow of protons and net - protons in mid - central collisions ( 10 - 40% centrality ) at midrapidity ( @xmath1 ) shows a clear non - monotonic variation with respect to @xmath11 ( @xmath6 ) .
the minimum value of @xmath1 lies somewhere between @xmath11 ( @xmath6 ) = 27 gev(160 mev ) to 11.5 gev(315 mev ) .
the net - proton @xmath1 changes sign twice in the beam energy range studied .
this observable which is driven by the pressure gradients developed in the system is sensitive to first order phase transition effects .
the energy dependence of the measured @xmath1 is consistent with a theoretical hydrodynamic model calculation with first order phase transition @xcite .
deviations of @xmath29@xmath34 and @xmath5 for net - proton distribution in 0 - 5% centrality is observed at @xmath24 = 19.6 and 27 gev from : ( a ) 70 - 80% peripheral collisions , ( b ) poisson and hadron resonance gas expectation value of close to unity and ( c ) transport model based urqmd calculation within the experimental acceptance .
the deviations of @xmath50@xmath51 and @xmath29@xmath34 below poisson expectation are qualitatively consistent with a qcd based model which includes a cp @xcite .
higher statistics data set at @xmath11 @xmath22 20 gev in the second phase of the beam energy scan program is needed to clarify whether the energy dependence of the observable will follow a non - monotonic variation with a minimum around @xmath11 = 27 gev to 11.5 gev , as observed for the net - proton @xmath1 , or a monotonic variation with @xmath11 .
99 i. arsene _ et al .
_ , nucl .
phys . a * 757 * , 1 ( 2005 ) ; b. b. back _ et al . _ , nucl .
phys . a * 757 * , 28 ( 2005 ) ; j. adams _ et al . _ , nucl . phys . a * 757 * , 102 ( 2005 ) ; k. adcox _ et al . _ , nucl . phys . a * 757 * , 184 ( 2005 ) ; s. chatrchyan _ et al . _ , jhep * 05 * , 063 ( 2012 ) ; b. abelev _ et al . _ , phys . rev .
lett . * 109 * , 072301 ( 2012 ) .
y. aoki _ et al .
_ , nature * 443 * , 675 ( 2006 ) .
m. luzum and p. romatschke , phys . rev .
c * 78 * ( 2008 ) 034915 [ erratum - ibid .
c * 79 * ( 2009 ) 039903 ] [ arxiv:0804.4015 [ nucl - th ] ] .
h. song , s. a. bass , u. heinz , t. hirano and c. shen , phys .
* 106 * ( 2011 ) 192301 [ erratum - ibid .
* 109 * ( 2012 ) 139904 ] [ arxiv:1011.2783 [ nucl - th ] ] .
i. vitev and m. gyulassy , phys .
rev . lett .
* 89 * ( 2002 ) 252301 [ hep - ph/0209161 ] .
s. a. bass , c. gale , a. majumder , c. nonaka , g. -y .
qin , t. renk and j. ruppert , phys .
c * 79 * ( 2009 ) 024901 [ arxiv:0808.0908 [ nucl - th ] ] .
k. fukushima and t. hatsuda , rept .
* 74 * ( 2011 ) 014001 [ arxiv:1005.4814 [ hep - ph ] ] .
d. h. rischke , y. pursun , j. a. maruhn , h. stoecker and w. greiner , heavy ion phys .
* 1 * ( 1995 ) 309 [ nucl - th/9505014 ] .
j. brachmann , s. soff , a. dumitru , h. stoecker , j. a. maruhn , w. greiner , l. v. bravina and d. h. rischke , phys .
c * 61 * ( 2000 ) 024909 [ nucl - th/9908010 ] .
h. stoecker , nucl .
a * 750 * ( 2005 ) 121 [ nucl - th/0406018 ] .
l. adamczyk _ et al . _ [ star collaboration ] , arxiv:1401.3043 [ nucl - ex ] .
v. koch , a. majumder and j. randrup , phys .
lett . *
95 * , 182301 ( 2005 ) ; m. asakawa , u. w. heinz and b. muller , phys .
lett . *
85 * , 2072 ( 2000 )
. m. a. stephanov , phys .
lett . * 102 * , 032301 ( 2009 ) ; c. athanasiou _ et al .
_ , phys .
d * 82 * , 074008 ( 2010 ) .
b. berdnikov and k. rajagopal , phys .
d * 61 * , 105017 ( 2000 ) .
s. ejiri , f. karsch and k. redlich , phys .
b * 633 * , 275 ( 2006 ) ; m. cheng _ et al .
_ , phys . rev . *
d 79 * , 074505 ( 2009 ) ; b. stokic _
et al . _ , phys .
* b 673 * , 192 ( 2009 ) ; s. gupta _ et al .
_ , science * 332 * , 1525 ( 2011 ) .
r. v. gavai and s. gupta , phys .
b * 696 * , 459 ( 2011 ) .
s. a. bass , m. belkacem , m. bleicher , m. brandstetter , l. bravina , c. ernst , l. gerland and m. hofmann _ et al .
. part .
* 41 * ( 1998 ) 255 [ prog
. part .
phys . * 41 * ( 1998 ) 225 ] [ nucl - th/9803035 ] .
m. bleicher , e. zabrodin , c. spieles , s. a. bass , c. ernst , s. soff , l. bravina and m. belkacem _ et al .
_ , j. phys .
g * 25 * ( 1999 ) 1859 [ hep - ph/9909407 ] .
m. isse , a. ohnishi , n. otuka , p. k. sahu and y. nara , phys .
c * 72 * ( 2005 ) 064908 [ nucl - th/0502058 ] .
m. m. aggarwal _ et al .
_ [ star collaboration ] , phys .
lett . * 105 * ( 2010 ) 022302 [ arxiv:1004.4959 [ nucl - ex ] ] .
l. adamczyk _ et al . _ [ star collaboration ] , phys .
* 112 * ( 2014 ) 032302 arxiv:1309.5681 [ nucl - ex ] .
f. karsch and k. redlich , phys
. lett .
b * 695 * , 136 ( 2011 ) .
p. garg , d. k. mishra , p. k. netrakanti , b. mohanty , a. k. mohanty , b. k. singh and n. xu , phys .
b * 726 * , 691 ( 2013 ) .
a. bzdak , v. koch and v. skokov , phys .
c * 87 * , 014901 ( 2013 ) .
x. luo , j. xu , b. mohanty and n. xu , j. phys .
g * 40 * , 105104 ( 2013 ) .
|
two measurements related to the proton and anti - proton production near midrapidity in @xmath0 = 7.7 , 11.5 , 19.6 , 27 , 39 , 62.4 and 200 gev au+au collisions using the star detector at the relativistic heavy ion collider ( rhic ) are discussed . at intermediate impact parameters
the slope parameter of the directed flow versus rapidity ( @xmath1 ) for the net - protons shows a non - monotonic variation as a function of the beam energy .
this non - monotonic variation is characterized by the presence of a minimum in @xmath1 between @xmath0 = 11.5 and 19.6 gev and a change in the sign of @xmath1 twice between @xmath2 = 7.7 and 39 gev . at small impact parameters the product of the moments of net - proton distribution , kurtosis @xmath3 variance ( @xmath4 ) and skewness @xmath3 standard deviation ( @xmath5 )
are observed to be significantly below the corresponding measurements at large impact parameter collisions for @xmath2 = 19.6 and 27 gev .
the @xmath4 and @xmath5 values at these beam energies deviate from the expectations from poisson statistics and that from a hadron resonance gas model .
both these measurements have implications towards the understanding of the qcd phase structures , the first order phase transition and the critical point in the high baryonic chemical potential region of the phase diagram .
|
with the release of the cosmic microwave background ( cmb ) maps obtained with the planck satellite , a string of properties of primordial scalar perturbations has been established with unprecedented accuracy . in particular , possible deviations from gaussianity and adiabaticity are now subject to quite severe constraints , while exact scale - invariance of the primordial power spectrum is excluded at more than @xmath1 c.l .
these observations show no departure from the simplest idea of single scalar slow roll inflation , while narrowing the window for many alternative scenarios . along with gaussianity and adiabaticity ,
large - scale statistical isotropy ( si ) of the universe is a basic prediction of standard inflationary cosmology .
this stems from the spin-0 nature of the inflaton and the isotropy of the ( quasi)de sitter space - time @xcite resulting in the rotational invariance of the inflaton field s correlation functions ; together with the isotropy of the background metric during radiation- and matter - dominated stages , this implies the si of cmb temperature fluctuations @xmath2 .
in particular , this means that the variance @xmath3 is independent of the direction @xmath4 in the sky .
this can be paraphrased in harmonic space as the diagonality of the covariance matrix .
although si is a robust prediction of inflation , there are examples of models which break this symmetry . at the level of primordial scalar perturbations @xmath5 , this amounts to saying that the power spectrum is direction - dependent , @xmath6 it is convenient to expand the function @xmath7 in a series of spherical harmonics @xcite , @xmath8 generically , the coefficients @xmath9 may have a dependence on the cosmological wavenumber @xmath10 .
depending on the type of directional - dependence , i.e. , the function @xmath7 , one can classify the anisotropies of the early universe as follows : * axisymmetric quadrupole . in this case
there exists a reference frame where all but one of the @xmath9 coefficients can be turned to zero , leaving only @xmath11 .
this type of statistical anisotropy ( sa ) is the most widely discussed in the literature . besides its simplicity , it is a common prediction of anisotropic models of inflation as well as of some alternatives .
the former include the historically first ackermann carroll wise model @xcite , scenarios with vector curvaton @xcite , scenarios with a gauge field coupled to waterfall fields in hybrid inflation @xcite and the generic class of models with a single maxwellian gauge field coupled to the inflaton @xcite .
alternatives to inflation are represented by models of the ( pseudo)conformal universe @xcite , i.e. , conformal rolling scenario @xcite and galilean genesis @xcite .
the common denominator of those scenarios is the existence of a _ single _ long - ranged vector , which is responsible for si breaking .
rotations with respect to this vector leave the primordial power spectrum intact .
thence , the axial symmetry of the quadrupole . * general quadrupole .
this type of prediction is less widely discussed in literature .
it follows , for example , from inflationary scenarios with _
maxwellian fields coupled to the inflaton @xcite as well as from the ( pseudo)conformal universe @xcite . in this case a single parameter is not enough to describe the anisotropy , as the axial symmetry is broken ; a second quantity the quadrupole shape @xmath12needs to be taken into account @xcite
. the parameter @xmath12 measures the deviation from axial symmetry .
the particular choice @xmath13 corresponds to leaving the latter intact .
we will sometimes refer to the shape @xmath12 as the angle for reasons which will become clear in section 2 .
sa of general quadrupolar form is the main focus of this work .
* higher multipoles .
this prediction about sa arises in some versions of the ( pseudo)conformal universe @xcite ; we will not deal with higher multipoles here .
+ we see how sa at least in principle could be a useful tool for discriminating among inflationary models as well as alternative frameworks . so far , most of the data analysis focussed on axisymmetric quadrupolar sa .
the bounds obtained by exploiting the quadratic maximum likelihood estimators @xcite are given by @xcite ( see also ref .
@xcite ) , @xmath14 for planck 2015 data and @xmath15 ( @xmath16 ) for wmap9 data @xcite .
planck collaboration extended the analysis so that to include the possible @xmath10-dependence of the amplitude @xmath0 @xcite .
in many cases , these constraints imply very stringent limits on the intrinsic parameters of the anisotropic early universe scenarios . in the present paper we search for the signatures of the general quadrupolar sa in the cosmological data for @xmath10-independent @xmath17 , see eq . , using planck 2015 maps .
following ref .
@xcite , we consider the data provided at @xmath18ghz and @xmath19ghz , and their cross - correlation .
when formulating the final constraints , we stick to the cross - correlated data as the cleanest probe of sa . the non - observation of any departures from si bounds the amplitude of the general quadrupole @xmath0 .
notably , the data demonstrate different level of agreement with different quadrupole shapes .
this is clear from the resulting constraints on @xmath0 , summarised in table [ table1 ] , as a function of the shape quantified by the parameter @xmath12 .
we see a slight , but not statistically significant , preference towards negative amplitudes @xmath0 .
that tendency , prominent for sufficiently small angles @xmath12 , vanishes at larger values of @xmath12 , see section 3 for discussions . for comparison purposes we also test the general quadrupole with planck 2013 data .
the latter , however , turn out to be insensitive to the shape of the quadrupole . as a result , the final constraints on the amplitude @xmath0 are independent on the angle @xmath12 .
these constraints on @xmath0 match very well the limits of ref .
@xcite ( deduced specifically for the axisymmetric quadrupole ) .
generically , in early universe models ( be it inflation or its competitors ) , the amplitude @xmath0 and the angle @xmath12 are not genuine model parameters , but they are random variables with distributions determined by the parameters specific for each scenario .
this is a common situation in anisotropic inflationary scenarios with vector fields and in the ( pseudo)conformal universe .
thus , a separate analysis is required in order to limit those models .
we perform this analysis in section 4 with a focus on inflationary scenarios which comprise ( a collection of ) maxwellian fields non - minimally coupled to the inflaton @xcite . in that case
, sa is sourced by the infrared modes of these maxwellian ( gauge ) vector fields , which follow a gaussian distribution with a dispersion that grows linearly with the duration of inflation ( in terms of the number of e - folds ) @xcite .
therefore , the number of e - folds can be constrained from the non - observation of sa . in section 4
, we also briefly revisit the limits on some versions of the ( pseudo)conformal universe .
this paper is organised as follows . in section [ sec : est ] , we discuss the parametrisation of sa . in section 3 , we assess the sensitivity of the planck data to the new parameter @xmath12 .
we constrain inflationary scenarios with multiple vector fields from the non - observation of sa in section 4 .
our constraining scheme there is independent on the results of sections 2 and 3 .
therefore , the reader interested only in those models may go directly to section 4 .
upon choosing an appropriate coordinate system , one can write a general quadrupole as @xcite @xmath20 \sin \chi \right\ } \ ; .\ ] ] here @xmath0 is the quadrupolar amplitude , and @xmath12 is an extra parameter ( angle ) which measures the departure from an axisymmetric quadrupole . notice that in the case @xmath21 we come back to the usual axial symmetry . the angles @xmath22 and @xmath23 correspond to the direction of the cosmological mode @xmath24 in the new coordinate system . the orthonormal basis of this coordinate system is given by the triad of vectors @xmath25 aligned with the axis @xmath26 , @xmath27 and @xmath28 , respectively . the representation of a general quadrupole has a particularly clear meaning in terms of multipole vectors @xcite . up to a constant factor , we write @xmath29 where @xmath30 and @xmath31 are mutipole vectors . as we show in the appendix , the multipole vector representation reduces to the form provided that one of the vectors is aligned with the @xmath32 axis of the new reference frame , while the other is lying in the @xmath33 plane . in the special case when two multipole vectors coincide , one recovers the axisymmetric quadrupole .
the freedom of choosing the reference frame does not introduce any ambiguity in the quantities @xmath0 and @xmath12 : they are uniquely defined in the region @xmath34 and @xmath35 , only @xmath36 is uniquely defined . this has no physical consequences since the difference between quadrupoles with positive and negative @xmath0 smoothly goes to @xmath37 as @xmath38 , see section [ ssec : results ] . ] . for practical purposes
it is convenient to express the coefficients @xmath17 in terms of the amplitude @xmath0 and the shape @xmath12 , @xmath39 where @xmath40 with the planck data at hand , one may reconstruct the coefficients @xmath17 and confront observations with the coefficients calculated for the given parameters @xmath0 and @xmath12 using eqs . and .
notice that although we are always considering two parameters ( viz , @xmath0 and @xmath12 ) , an observer looking for imprints in the cmb will have to deal with three additional numbers determining the actual orientation of the quadrupole in the sky ( namely , three angles fixing the orthornomal vectors @xmath41 and @xmath42 ) .
however , the theory typically tells nothing regarding the directions of these vectors , which should thus be drawn from uniform distributions .
marginalising over these three extra parameters one is left with the amplitude @xmath0 and @xmath12 alone .
conventionally , one employs quadratic maximum likelihood ( qml ) estimators to derive the @xmath9 coefficients from cmb maps . originally designed in ref .
@xcite , they yielded the most stringent limits on sa to date @xcite .
we slightly modify the qml estimators to make them appropriate for the cross - correlation analysis @xcite .
that is , we consider estimators of the form @xmath43 where @xmath44 here @xmath45 are related to standard cmb temperature coefficients @xmath46 by @xmath47 the upperscripts @xmath48 and @xmath49 denote a particular frequency band ( @xmath18ghz or @xmath19ghz in our case ) ; @xmath50 is a covariance ( including the noise ) , which corresponds to _ statistically isotropic _ cosmological signal .
the coefficients @xmath51 in eq . are given by @xmath52 where @xmath53 is an isotropic primordial power spectrum and @xmath54 is a transfer function .
the structure constants @xmath55 are expressed via the wigner 3j - symbols @xmath56 the estimators are normalised by the fisher matrix @xmath57 given by @xmath58 in the homogeneous noise and full sky approximations , the fisher matrix can be written in analytic form @xcite as @xmath59 here @xmath60 , where @xmath61 and @xmath62 are the primordial and noise angular spectra derived from the @xmath48th band , respectively ; @xmath63 for @xmath64 and @xmath65 otherwise .
given an incomplete sky coverage , we proceed with a slight modification of the fisher matrix , @xmath66 where @xmath67 is the unmasked fraction of the sky .
we followed the steps described below to derive @xmath9 coefficients from the data and from monte carlo ( mc ) simulated maps : * we obtain the temperature coefficients @xmath68 from planck 2015 data corresponding to 29 months of high frequency instrument ( hfi ) observations at frequencies @xmath18ghz and @xmath19ghz @xcite .
we employ the hfi gal40 galactic plane mask ( hfi_mask_galplane - apo0_2048_r2.00.fits ) and hfi point sources mask ( hfi_mask_pointsrc_2048_r2.00.fits ) .
the unmasked fraction of the sky is @xmath69 . * from @xmath68
, we calculate the @xmath70 coefficients defined by eq . .
following ref .
@xcite , we employ multigrid preconditioners
@xcite at this step to reduce computational cost . * the @xmath51 coefficients given by eq
. are evaluated with the _ camb _ package @xcite .
* we compute the wigner 3j - symbols and provide summations in eqs . and using _ gsl _
@xcite and _ slatec _ @xcite libraries .
summations are cut at the multipole number @xmath71 . for @xmath72 ,
the observed signal is dominated by instrumental noise . * finally , averaging in eq .
is performed by repeating the procedure for 100 monte - carlo generated statistically isotropic maps .
the latter are constructed using full focal plane simulations for cmb and noise maps @xcite coadded with the _ smica _ foreground map ( hfi_compmap_foregrounds - smica_2048_r2.00.fits ) @xcite . to assess the sensitivity of the planck data to sa , we start with the @xmath73-statistics defined by @xmath74 ( the upperscript q here serves to distinguish the @xmath73 coefficients from the amplitudes @xmath75 describing the angular power spectrum ) .
this has been used in refs .
@xcite to constrain models of the ( pseudo)conformal universe and anisotropic inflationary scenarios with one vector field .
the reconstructed @xmath73 coefficients are shown in fig . [ 143 ] .
for the sake of completeness , we consider also higher multipoles of sa up to @xmath76 .
the data at frequency @xmath18ghz is consistent with the hypothesis of si ( within @xmath77 c.l . ) . at the same time
, the data at @xmath19ghz exhibits a quadrupolar sa .
we attribute this departure from si to unknown systematic effects .
these are typically uncorrelated between different frequency bands , and thus are mitigated upon cross - correlating the data .
it is indeed the case , as is clearly seen from fig .
a similar observation was made with the planck 2013 dataset @xcite . in that case , however , the signal of ( non - cosmological ) sa was identified at @xmath78 . once again , the consistency with si was restored in the cross - correlated data . in what follows ,
we stick to the latter as the one least plagued by possible systematic effects . in view of our objectives ,
the @xmath73-statistics is not enough , since the @xmath79 coefficients are insensitive to the possible @xmath12-dependence of a general quadrupole .
drops out .
] we find that the strategy adopted in ref .
@xcite is more appropriate for our purposes here . given that the estimator @xmath80 is affected by a large number of random quantities , including noise realisation and random correlations of cmb signal and foregrounds , one may consider a gaussian likelihood function , @xmath81^{\dagger}{\bf f}^{-1 } \left[\hat{{\bf q}}- { \bf q } ( g _ * , \chi , { \bf a } , { \bf c } ) \right ] \right ) \ ; .\ ] ] substituting eqs . and into eq .
, one obtains a posterior distribution for @xmath0 , @xmath12 , and the vectors @xmath41 and @xmath42 .
we then marginalise over these @xmath41 and @xmath42 , and find the joint distribution of @xmath0 and @xmath12 : @xmath82 .
the result is shown in fig .
[ likelihood ] .
this distribution is not gaussian because the parameters @xmath12 , @xmath41 , and @xmath42 enter non - linearly in @xmath17 , see eq . .
we see from fig .
[ likelihood ] that the data treats different values of the parameters @xmath12 non - uniformly . for small @xmath12 , when the axial symmetry is restored , the data prefers negative @xmath0 . for larger @xmath12 , the preference shifts towards positive values of @xmath0 .
notably , in the limit @xmath83 , the distribution becomes symmetric with respect to a change of the sign of @xmath0 .
this is especially clear from the distributions of fig .
[ likelihoods ] obtained from @xmath82 of fig . [ likelihood ] by cutting the latter at a given value of the angle @xmath12 ( we choose @xmath84 ) . in fig .
[ likelihoods ] we also plot the likelihood marginalised over possible values of the angle @xmath12 . at this level , we assume a homogeneous distribution for @xmath12 .
the final limits on the amplitude @xmath0 are summarised in table [ table1 ] .
the physical reason for the symmetrisation of the likelihood at large values of the angle @xmath12 is as follows . in the case
@xmath85 , eq . reduces to @xmath86 where we omitted an irrelevant constant factor .
the r.h.s . here is symmetric under the coordinate transformation @xmath87 , supplemented by a change of sign for the amplitude @xmath0 , i.e. , @xmath88 .
therefore , upon marginalising over the directions of the vectors @xmath41 and @xmath42 , we get @xmath89 .
the two cases are therefore physically indistinguishable . .
[ cols="^,^,^",options="header " , ] now let us compare the planck 2013 and 2015 datasets .
the joint likelihood @xmath82 obtained from planck 2013 data is shown in fig .
[ likelihoods2013 ] .
notably , this distribution corresponds to nearly zero best - fit value of the amplitude @xmath0 .
the reason is that the coefficients @xmath90 reconstructed from planck 2013 maps are very close ( i.e. , well within @xmath91 interval ) to the average value obtained from statistically isotropic monte - carlo maps . as a result ,
the distribution @xmath82 is independent on the angle @xmath12evidently , for vanishing sa the data can not discriminate between different quadrupole shapes .
our final limits on the amplitude @xmath0 are shown in table [ table1 ] .
they agree very well with the limits deduced in ref .
@xcite for the case of the axisymmetric quadrupole .
typically , in early universe scenarios , the amplitude @xmath0 and the angle @xmath12 are random variables with statistical properties determined by the intrinsic model parameters .
that is , there is no one - to - one correspondence between the quadrupolar amplitude and shape and the theoretical parameters in the lagrangian . in this situation
, the relevant constraints can not be immediately inferred from those of table [ table1 ] .
we perform such analysis in this section for inflationary scenarios in which several maxwellian spectator fields are non - minimally coupled to an inflaton @xcite .
a similar analysis for the case of the ( pseudo)conformal universe ( but with planck 2013 data ) can be found in ref .
@xcite : we briefly revisit those limits with the new dataset at the end of the section .
one way to achieve sa in inflation is , by virtue of their directional nature , to introduce vector fields .
the most well - known example is the model where the maxwellian fields are coupled to the inflaton itself @xcite . in that case
the vectors @xmath92 gauge invariance is preserved , and one has a chance to achieve sa without developing catastrophic ghost instabilities @xcite . in the literature ,
the scenario with a single gauge field is the most popular one . in this case ,
one deals with a directional dependence in the scalar perturbations power spectrum which is axisymmetric and characterised by a _ negative _ amplitude
@xmath0 @xcite . in the setup with multiple gauge fields @xcite the axial symmetry
is broken and the resulting sa is a general quadrupole .
we consider the gauge sector action : @xmath93 here the subscript @xmath94 runs over the collection of @xmath95 gauge fields @xmath96 for which @xmath97 is the strength . for simplicity we assumed that the coupling of the gauge fields to the inflaton is universal , but generalisations are straightforward and do not significantly change our results .
if we were to take a trivial kinetic gauge function , @xmath98 , the contribution of each gauge field would redshift away adiabatically due to the expansion of the universe .
this decay can be prevented by choosing appropriate dynamics for @xmath99 .
one well justified possibility is @xmath100 @xcite ; this might in fact be the only sensible option since it is a quite generic attractor of this system @xcite . in that case , the contribution of the `` electric '' energy density remains constant during inflation also an attractor for an inverted coupling function @xmath101would result into a `` magnetic '' energy density which is constant with time ; this option is typically invoked in the context of magnetogenesis @xcite . ] .
this makes it possible to generate non - trivial sa described by @xcite @xmath102 here @xmath103 , is the classical `` electric '' field .
see the discussion below on its origin .
the amplitudes @xmath104 are given by @xmath105 where @xmath106 is the number of e - folds between horizon crossing of mode @xmath10 and the end of inflation ; @xmath107 is the standard slow roll parameter and @xmath108 is the inflaton potential .
the amplitudes @xmath109 here are not to be confused with the amplitude @xmath0 of the general quadrupole defined from eq . .
while the former are always negative , the latter is allowed to take on positive values as well .
the `` electric '' fields @xmath103 have two sources .
one is purely classical , corresponding to the attractor solution of the background equations of motion @xcite .
the second is due to quantum fluctuations which get enhanced and stretched during inflation ( and finally classicalise once they leave the horizon ) @xcite .
this is an infrared component which is built up of all modes , processed by inflation , which are now beyond our observable horizon .
one thus writes @xmath103 as follows @xmath110 here @xmath111 is the status of the infrared vector at the time when the gauge mode matching the size of our observable universe crossed the horizon , @xmath112 ; here @xmath113 denotes the present hubble rate in conformal time units .
modes crossing the horizon after @xmath114 are ignored because they look inhomogeneous from our point of view and hence do not contribute significantly to the global asymmetry parametrised by @xmath0 .
see ref .
@xcite for a careful discussion of this point .
we model the vector as a gaussian field with zero mean and variance @xcite @xmath115 where @xmath116 is the `` extra '' e - folds of inflation , namely , the number of e - folds between start of inflation and @xmath114 .
as far as the constraining procedure is concerned , barring cancellations , our limits would be conservative , as we attribute all anisotropy to one component only ; any other additional anisotropy would only exacerbate the discrepancy with the data .
notice furthermore that in multivector scenarios background isotropy is attractive , and the anisotropy is then expected to only arise due to the infrared fluctuations .
hence , we will be interested in the case for which the purely classical `` electric '' fields @xmath117 give a negligible contribution to sa , i.e. , @xmath118 .
then , the `` extra '' number of e - folds @xmath119 is the only relevant parameter which affects si and generates sa .
before diving into the routine of the constraining procedure , let us make a short remark .
by glancing at eq .
, one may see that the amplitude @xmath0 is scale - dependent . despite this dependence being relatively mild
, it may essentially bias our constraints , because planck has access to a wide range of multipoles .
in reality , the dependence present in @xmath106 and the one due to the slow roll potential @xmath120 compensate each other with a high accuracy : @xmath121 substituting the experimental central value @xmath122 and @xmath123 , we observe that two values on the r.h.s .
approximately sum up to zero .
thus , we can safely set @xmath124 in eq . .
it is convenient to rewrite the amplitudes @xmath109 given by eq .
as follows @xmath125 where @xmath126 are gaussian vectors collinear with @xmath127 , characterised by zero means and variances @xmath128 note that each component ( @xmath129 ) has the same variance since the background spacetime is rotationally symmetric to very good accuracy .
here we made use of the slow roll relation @xmath130 in order to get rid of the dependence on the potential @xmath131 and the slow - roll parameter @xmath107 .
to constrain the number @xmath119 we use the following strategy : starting from a given value of @xmath119 we generate an ensemble of @xmath132 sets of gaussian vectors @xmath126 from eq .
( [ gauzz ] ) . for each set
, we calculate the coefficients @xmath17 defined by the relation and compute the likelihood of eq . using planck data .
the latter is then averaged over the ensemble of sets .
we run this algorithm for different numbers of maxwellian fields : @xmath133 $ ] in steps of 1 and different @xmath119 values .
the results are presented in fig .
notice that the best fit value of @xmath119 falls roughly as @xmath134 with the number of fields , in accordance with theoretical expectations @xcite . for comparison purposes ,
the constraint on the number of e - folds @xmath119 in the case of the single vector field read : @xmath135 ( @xmath136 at @xmath137 cl for @xmath138 ) .
this limit is only a very moderate improvement compared to the analogous wmap9 limit of ref .
@xcite .
[ [ the - pseudoconformal - universe - revisited . ] ] the ( pseudo)conformal universe revisited .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + before closing , let us comment on some immediate implications of eq . for the ( pseudo)conformal universe @xcite .
the latter is an alternative to inflation , which attributes the approximate flatness of primordial peturbations to the assumed conformal symmetry of the early universe @xcite . as far as sa is concerned ,
the ( pseudo)conformal universe is practically equivalent ( modulo a constant factor ) to inflation augmented with a single maxwellian field @xcite with the wavenumber @xmath10 .
hence , it makes a negligible imprint on the cmb . ] .
this is true at least for the subclass of ( pseudo)conformal universe models in which the cosmological perturbations of interest remain frozen after the conformal phase and before the hot stage / reheating @xcite ( sub - scenario a in the notation of ref .
the constraint of eq .
translates to @xmath139 here @xmath140 is the parameter which governs the non - trivial evolution in the ( pseudo)conformal universe ; @xmath141 is the hubble rate today , and @xmath142 is the cutoff on the modes feeding into sa , see ref .
@xcite for details .
the comparison of planck data with the predictions of multivector inflation models is supported by the russian science foundation grant no .
14 - 12 - 01430 ( g.r . ) .
the analysis of the ( pseudo)conformal universe models is supported by the russian science foundation grant 14 - 22 - 00161 ( g.r . ) .
is supported by the erc grant put808 and the erdf coe program .
instead of the quantities @xmath0 , @xmath12 , @xmath41 and @xmath42 , one may wish to work in terms of the multipole vectors . in order to do so , we follow ref .
@xcite and write down the multipole vector definition for a quadrupole : @xmath143 where @xmath144 and @xmath145 are two multipole vectors assumed to be normalised as @xmath146 the constant @xmath147 can be obtained by integrating the left and the right hand sides of eq . over the directions of the cosmological mode @xmath24 .
the result reads @xmath148 notice that the relation does not fix the signs of the multipole vectors and of the amplitude @xmath149 .
we partially eliminate this ambiguity by requiring , with no loss of generality , @xmath150 the remaining freedom which is due to the simultaneous change of sign of two multipole vectors does not affect any physical quantity , since the vectors always enter only in bilinear combinations . to obtain the general quadrupole in the form
we choose a coordinate system with the @xmath32 axis aligned with one of the multipole vectors . clearly , we can organise this in two possible ways accordingly to the number of multipole vectors . for concreteness ,
we pick the vector @xmath151 , i.e. , @xmath152 and @xmath153 . from the condition
, it follows that @xmath154 , and @xmath155 finally , we can rotate the coordinate system in such a way that the vector @xmath30 lies in the @xmath33 plane , i.e. , @xmath156 . to fix the coordinate system
completely , we require that @xmath157 . from eq . , we then obtain the quadrupole term in the form , @xmath158 \right\ } \ ; .\ ] ] now , we observe that there is a one - to - one correspondence between this parametrisation ( with the coordinate system fixed as discussed above ) and that of eq . in the region
@xmath159 and @xmath160 . to make it even clearer , we can obtain the explicit relations between @xmath0 and @xmath12 and the multipole vectors as @xmath161}}{2({\bf v}^{(2,1 ) } \cdot { \bf v}^{(2,2 ) } ) } \ ; , \ ] ] @xmath162 and @xmath163 one final remark is in order here . as it follows from the discussion above ,
there are three more coordinate system where the quadrupole takes the form .
one is associated with the interchange of the multipole vectors @xmath164 , while the other two are obtained from the latter two by rotating the @xmath165 plane , resulting into the simultaneous change of the sign of the multipole vectors @xmath166 .
clearly , this does not introduce any ambiguity in the definition of the amplitude @xmath0 and the angle @xmath12 ( still , we allow the latter to vary within the region @xmath167 ) .
this readily follows from expressions and , which are invariant under the interchange of the multipole vectors , i.e. , @xmath168 , and under the simultaneous change of their signs .
l. ackerman , s. m. carroll and m. b. wise , phys .
d * 75 * ( 2007 ) 083502 [ erratum - ibid .
d * 80 * ( 2009 ) 069901 ] [ astro - ph/0701357 ] .
k. dimopoulos , m. karciauskas , d. h. lyth and y. rodriguez , jcap * 0905 * ( 2009 ) 013 [ arxiv:0809.1055 [ astro - ph ] ] .
k. dimopoulos , m. karciauskas and j. m. wagstaff , phys .
b * 683 * ( 2010 ) 298 [ arxiv:0909.0475 [ hep - ph ] ] .
+ k. dimopoulos , int .
j. mod .
d * 21 * ( 2012 ) 1250023 [ erratum - ibid .
d * 21 * ( 2012 ) 1292003 ] [ arxiv:1107.2779 [ hep - ph ] ] .
s. yokoyama and j. soda , jcap * 0808 * ( 2008 ) 005 [ arxiv:0805.4265 [ astro - ph ] ] .
m. -a .
watanabe , s. kanno and j. soda , phys .
* 102 * ( 2009 ) 191302 [ arxiv:0902.2833 [ hep - th ] ] . m. -a
. watanabe , s. kanno and j. soda , prog .
* 123 * ( 2010 ) 1041 [ arxiv:1003.0056 [ astro-ph.co ] ] .
j. soda , class .
* 29 * ( 2012 ) 083001 [ arxiv:1201.6434 [ hep - th ] ] .
n. bartolo , s. matarrese , m. peloso and a. ricciardone , phys .
d * 87 * ( 2013 ) 023504 [ arxiv:1210.3257 [ astro-ph.co ] ] .
d. h. lyth and m. karciauskas , jcap * 1305 * ( 2013 ) 011 [ arxiv:1302.7304 [ astro-ph.co ] ] .
k. hinterbichler and j. khoury , jcap * 1204 * ( 2012 ) 023 [ arxiv:1106.1428 [ hep - th ] ] .
p. creminelli , a. joyce , j. khoury and m. simonovic , jcap * 1304 * ( 2013 ) 020 [ arxiv:1212.3329 [ hep - th ] ] .
m. libanov , v. rubakov and g. rubtsov , jetp lett .
* 102 * ( 2015 ) no.8 , 561 [ pisma zh .
* 102 * ( 2015 ) no.8 , 630 ] ; [ arxiv:1508.07728 [ hep - th ] ] .
v. a. rubakov , jcap * 0909 * ( 2009 ) 030 [ arxiv:0906.3693 [ hep - th ] ] .
m. libanov and v. rubakov , jcap * 1011 * ( 2010 ) 045 [ arxiv:1007.4949 [ hep - th ] ] .
m. libanov , s. mironov and v. rubakov , `` non - gaussianity of scalar perturbations generated by conformal mechanisms , '' phys .
d * 84 * ( 2011 ) 083502 [ arxiv:1105.6230 [ astro-ph.co ] ] .
k. yamamoto , m. -a .
watanabe and j. soda , class .
* 29 * ( 2012 ) 145008 [ arxiv:1201.5309 [ hep - th ] ] .
m. thorsrud , f. r. urban and d. f. mota , jcap * 1404 * ( 2014 ) 010 [ arxiv:1312.7491 [ astro-ph.co ] ] .
m. thorsrud , d. f. mota and f. r. urban , phys .
b * 733 * ( 2014 ) 140 doi:10.1016/j.physletb.2014.04.028 [ arxiv:1311.3302 [ astro-ph.co ] ] .
m. libanov , s. ramazanov and v. rubakov , jcap * 1106 * ( 2011 ) 010 [ arxiv:1102.1390 [ hep - th ] ] .
r. adam _ et al . _
[ planck collaboration ] , astron .
astrophys . * 594 * , a8 ( 2016 ) doi:10.1051/0004 - 6361/201525820 [ arxiv:1502.01587 [ astro-ph.co ] ] .
a. lewis , a. challinor and a. lasenby , astrophys .
j. * 538 * ( 2000 ) 473 [ astro - ph/9911177 ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , astron .
astrophys .
* 594 * , a12 ( 2016 ) doi:10.1051/0004 - 6361/201527103 [ arxiv:1509.06348 [ astro-ph.co ] ] .
r. adam _ et al . _
[ planck collaboration ] , astron .
astrophys . * 594 * , a10 ( 2016 ) doi:10.1051/0004 - 6361/201525967 [ arxiv:1502.01588 [ astro-ph.co ] ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , astron . astrophys .
* 594 * ( 2016 ) a16 doi:10.1051/0004 - 6361/201526681 [ arxiv:1506.07135 [ astro-ph.co ] ] .
s. r. ramazanov and g. rubtsov , phys .
d * 89 * ( 2014 ) 043517 [ arxiv:1311.3272 [ astro-ph.co ] ] .
n. e. groeneboom and h. k. eriksen , astrophys . j. * 690 * ( 2009 ) 1807 [ arxiv:0807.2242 [ astro - ph ] ] .
n. e. groeneboom , l. ackerman , i. k. wehus and h. k. eriksen , astrophys . j. * 722 * ( 2010 ) 452 [ arxiv:0911.0150 [ astro-ph.co ] ] .
s. das , s. mitra , a. rotti , n. pant and t. souradeep , astron .
astrophys .
* 591 * ( 2016 ) a97 ; [ arxiv:1401.7757 [ astro-ph.co ] ] .
g. i. rubtsov and s. r. ramazanov , phys .
d * 91 * ( 2015 ) no.4 , 043514 doi:10.1103/physrevd.91.043514 [ arxiv:1406.7722 [ astro-ph.co ] ] .
c. j. copi , d. huterer and g. d. starkman , phys .
d * 70 * ( 2004 ) 043515 [ astro - ph/0310511 ] .
k. m. smith , o. zahn and o. dore , phys .
d * 76 * ( 2007 ) 043510 [ arxiv:0705.3980 [ astro - ph ] ] .
a. e. gumrukcuoglu , b. himmetoglu and m. peloso , phys .
d * 81 * ( 2010 ) 063528 [ arxiv:1001.4088 [ astro-ph.co ] ] .
+ j. ohashi , j. soda and s. tsujikawa , jcap * 1312 * , 009 ( 2013 ) doi:10.1088/1475 - 7516/2013/12/009 [ arxiv:1308.4488 [ astro-ph.co ] , arxiv:1308.4488 ] .
j. m. wagstaff and k. dimopoulos , phys .
d * 83 * , 023523 ( 2011 ) ; doi:10.1103/physrevd.83.023523 [ arxiv:1011.2517 [ hep - ph ] ] . s. hervik , d. f. mota and m. thorsrud , jhep * 1111 * ( 2011 ) 146 [ arxiv:1109.3456 [ gr - qc ] ] . m. thorsrud , d. f. mota and s. hervik , jhep * 1210 * , 066 ( 2012 ) ; doi:10.1007/jhep10(2012)066 [ arxiv:1205.6261 [ hep - th ] ] . a. ito and j. soda , phys . rev .
d * 92 * , no .
12 , 123533 ( 2015 ) ; doi:10.1103/physrevd.92.123533 [ arxiv:1506.02450 [ hep - th ] ] .
|
several early universe scenarios predict a direction - dependent spectrum of primordial curvature perturbations .
this translates into the violation of the statistical isotropy of cosmic microwave background radiation .
previous searches for statistical anisotropy mainly focussed on a quadrupolar direction - dependence characterised by a _
single _ multipole vector and an overall amplitude @xmath0 .
generically , however , the quadrupole has a more complicated geometry described by _ two _ multipole vectors and @xmath0 .
this is the subject of the present work . in particular , we limit the amplitude @xmath0 for different shapes of the quadrupole by making use of planck 2015 maps . we also constrain certain inflationary scenarios which predict this kind of more general quadrupolar statistical anisotropy .
inr - th/2016 - 047
|
it is commonly assumed that brown dwarfs are formed the same way as stars , except under conditions that lead to stellar objects with very small masses , i.e. from clouds that are very small , very dense , and very cold .
however , with the growing realization that brown dwarfs may be as common as stars , it is becoming disturbing that such special physical conditions are not readily found in the molecular clouds of our galaxy ( although they may exist elsewhere , see elmegreen 1999 ) .
alternatively , lin et al .
( 1998 ) suggested that the encounter between two protostars with massive disks could fling out tidal filaments with lengths of about 1000 au , out of which a brown dwarf might form .
however , the fact that brown dwarfs are increasingly discovered also in loose t tauri associations like taurus , where such encounters should be extremely rare , suggests that this mechanism is unlikely to be a major source of brown dwarfs . taking another approach ,
we have recently proposed that brown dwarfs have such extremely low masses because they were ejected from small clusters of nascent stellar embryos ( reipurth & clarke 2001 ) .
this can occur because the timescale for dynamical interactions and ejection is comparable to the timescale for collapse and build - up of a star . in the following ,
we outline the basic aspects of this model , and summarize the status of current efforts to numerically model such ejections .
observations over the last decade have established that young t tauri stars have the same or a slightly higher binary frequency than at the main sequence ( e.g. reipurth & zinnecker 1993 ; ghez et al .
1993 ; khler & leinert 1998 ) . a small , although not very well determined , fraction of both young and more evolved stars are also triple or higher - order multiple systems .
the situation is much less clear among the very youngest , still embedded stars , due to the difficulties of probing into the heavily shrouded environment of such objects .
however , new high resolution infrared techniques from the ground and space , as well as centimeter interferometry with e.g. the vla , are beginning to yield results . in a detailed study of 14 driving sources of giant herbig - haro ( hh ) flows , reipurth ( 2000 ) found that more than 80% are binaries , and of these half are higher order systems .
it should be noted that these are the actually observed frequencies , without corrections for the considerable incompleteness of the observations , and so the results are in fact consistent with the possibility that _ all _ giant hh flow sources may be binary or multiple systems .
these embedded outflow sources are of the order of 10@xmath1 yr old or less , and it follows that some of these systems must decay to reach the lower observed frequencies at later evolutionary stages . it is well established that non - hierarchical triple systems undergo rapid dynamical evolution and evolve into either a binary with a distant companion , i.e. a hierarchical triple system , or into a binary and an unbound , escaping third member ( see section 3 ) .
the binary system that is formed in this dynamical process is highly eccentric , and given that the triple disintegration is likely to take place while the stars are still actively accreting gas from an infalling envelope , it follows that the circumstellar disks will interact on an orbital timescale , which will lead to shrinkage of the orbit ( e.g. artymowicz & lubow 1996 ) .
these interactions are again likely to cause cyclic variations in the accretion rate , with consequent pulses in the outflow production , and the giant hh flows may therefore represent a fossil record of the birth and early evolution of binary systems ( reipurth 2000 ) .
altogether , it appears that the generation of giant hh flows , the birth of a binary , and the formation of brown dwarfs may all be different aspects of a single event , namely the dissolution of a small multiple system .
if the dominant mode of star formation involves the splitting of a core in to @xmath2 or more fragments then important new ingredients enter the dynamics which are not encountered in the cases @xmath3 or @xmath4 .
the reason for this is simply that most @xmath5 body configurations are dynamically unstable ; over several dynamical timescales the components exchange energy through the action of gravitational forces until they attain a stable equilibrium . in practice
, this generally means that the mini - cluster disintegrates , the energy for this process being derived from the formation of at least one binary system . after a number of dynamical timescales ,
therefore , the core no longer consists of a bound stellar system , but of a mixture of binary and single stars that drift apart so as to mingle with the ambient stellar field .
below we summarise the progress that has been made to date in quantifying such behaviour .
we would like to know what is the mass spectrum of the stars ( and brown dwarfs ) that are produced in this way , what are their binary statistics , how much circumstellar material young stars can bring with them as they emerge from such a core ?
we would also like to know what are the kinematic properties of the resulting stars and brown dwarfs .
unfortunately , the answers to these questions can be most readily obtained in the case of simulations that omit a vital ingredient ( i.e. gas ) ; and it is only recently that it has been possible to address these issues in hydrodynamical simulations .
even so , the twin demands of simulations that are both well resolved and can be performed a large number of times ( in order to obtain statistically meaningful results ) takes us close to the limits of what is computationally viable at the present time .
here we describe the results of simulations that are progressively more realistic , starting with pure n - body experiments and ending with what represents the current state of the art - gas dynamical simulations starting from turbulent initial conditions .
although these results will give a flavour of the physical processes involved , it should be stressed that we are not yet at a stage where statistical results can be derived from the most realistic experiments .
this inevitably means that the observational predictions ( described in section 4 ) remain somewhat provisional . the simplest possibility conceptually ( which is however not likely to be even approximately true in practice ) is if all the gas is instantly accreted on to each protostar so that the ensemble evolves thereafter as a system of point masses
this situation is straightforward to model as an nbody system and has been analysed by many authors ( e.g. van albada 1968 ; sterzik & durisen 1998 ) .
the usual outcome of such simulations is that the two most massive members of the system form a binary , whereas the remainder are ejected as single stars .
the ejection velocity of stars is related to their orbital velocity during a close three body encounter .
for example , in sterzik and durisen s simulations , the typical separation between stars is initially around @xmath6 au , and typical ejection velocities of single stars are 3 - 4 km s@xmath0 .
if the same system of point masses had been set up with separations a factor @xmath7 greater ( say ) , the resulting ejection velocities would be a factor @xmath8 smaller .
sterzik and durisen also quantified the dependence of ejection velocity on stellar mass , and found that for @xmath9 , such a dependence is very weak ; they however found a significant difference in the final velocities of single stars and binaries , with the centre of mass velocity of binaries being typically a factor @xmath10 less than that of singles .
the above simulations have all the advantages of computational simplicity and yield well defined predictions for the fraction of stars of various masses that end up in binaries ( mcdonald & clarke 1993 ) .
they however take no account of the fact that the fragments will not interact as point masses but will instead have their interactions mediated by the presence of circumstellar disks .
this situation is less straightforward to model numerically , since there are well known numerical difficulties in maintaining disks for many internal orbital timescales against the dispersive action of viscosity .
some insight into the expected role of disks can be obtained from the calculations of mcdonald & clarke ( 1995 ) , who included the effect of star - disk interactions in a parameterised form ( see also clarke & pringle 1991 ) .
the main role of disks is to harden temporary binaries so as to protect them against disruptive encounters with other cluster members . as a result ,
more than one binary can be formed in each cluster ; although the most massive cluster member is always in a binary , its companion is now picked at random from the cluster members .
thus the net effect of star - disk interactions is to boost the numbers of lower mass stars that end up in binaries ( either as primaries or secondaries ) relative to the dissipationless ( nbody ) case .
however , neither of the sorts of simulations described above can say anything about the stellar initial mass function , since stellar masses are assigned at the outset of the simulation .
such instantaneous mass assignment is of course a very poor approximation to the behaviour of real fragmenting cores : the interactions that lead to the formation of binaries and break up of the cluster occur over a few dynamical times , but the infall of mass onto each of the stars happens on a comparable timescale .
thus realistic simulations need to address the whole process as a hydrodynamic one and follow through the evolution of a core that is initially @xmath11 gas . to date
, some progress has been made on this issue by setting up cores in which ` seeds ' of collapsed gas have been planted ( bonnell et al 1997 ; bonnell et al .
these seeds grow in mass due to gas accretion in an inequitable manner - their orbital histories determine whether they spend much time in the densest central regions of the core and hence how much mass they acquire .
such simulations vividly demonstrate how ` competitive accretion ' works : seeds that get an early headstart in the race for mass tend to settle into the cluster core and thereby acquire more mass , whereas seeds that do not grow much initially are more likely to be flung out of the core and hence be prevented from further growth .
thus competitive accretion provides a ready mechanism for obtaining a large dynamic range of final stellar masses from arbitrary initial conditions .
recent simulations by delgado - donate ( in preparation ) have begun to quantify the imf produced by competitive accretion during the break up of small clusters ( figure 1 ) and find that it is broadly compatible with the observed imf .
( note that in these simulations no disks are formed around the protostars due to the absence of small scale turbulence in the initial core .
thus although these simulations model the accretion of gas on to the stars as they form , and the consequent shrinkage of the protostellar separations , they do not model the effects of star - disk interactions ) . the dynamical outcome of such disintegrations shares many qualitative similarities with the dissipationless ( nbody ) results of sterzik and durisen . as in their simulations
, there is no appreciable dependence of final velocity on resulting stellar mass ( see figure 2 ) , but the binaries attain speeds that are a factor @xmath12 less than the typical ejection speeds of the single stars . one notable difference , however , is the relationship between the typical ejection speeds and the initial parameters of the core . in the dissipationless simulations , the final velocities of ejected stars are of the same order as the virial velocity of the initial core . in the gas dynamical simulations , there are also many objects that are ejected early on with such velocities , as shown in figure 2 , where the initial virial velocity of the parent core is around @xmath13 km s@xmath0 .
however , figure 2 also shows that many stars are ejected with velocities much greater than this : about half the stars attain velocities greater than @xmath14 times the initial virial velocity ( the corresponding number for sterzik and durisen s simulations is less than @xmath15 ) and there is also a significant minority that attain velocities more than a factor @xmath7 greater than the initial virial value .
this may be readily understood inasmuch as in the gas dynamical simulations , the systems shrink as the gas is accreted , so that later interactions are much closer and produce correspondingly larger velocity kicks .
thus in these gas dynamical simulations ( where the typical initial separations of protostars are @xmath16 au ) , the ejection velocities are similar to sterzik and durisen s dissipationless calculations in which the stars are initially separated by only of order @xmath6 au .
these simulations also give some insight into the binary pairing characteristics of the resulting stars .
the outcome of each five star simulation is the production of a binary ( sometimes a triple ) .
( note that these simulations do not give reliable results for the separations of the resultant binaries , since they do not include star - disk interactions , which will tend to harden the binaries considerably ) .
it is notable that these simulations produce extreme mass ratio pairs ( e.g. @xmath17 ) quite readily , which contrasts strongly with pure nbody results , where such an outcome is almost unknown .
a further step towards reality can be obtained by abandoning the artificial distinction between seeds and smoothly distributed background gas in the above simulations .
a more physical approach is to start with gas that is subject to a supersonic turbulent velocity field which rapidly generates a richly non - linear density structure in the gas ( klessen et al .
2000 ; klessen 2001 ) .
such simulations follow not only the competitive accretion between contending ` stars ' and their dynamical interactions , but also the very formation of the stars from pockets of jeans unstable gas .
the most ambitious simulation to date is that of bate ( see bate et al .
2001 ) which models a system that will form of the order of a hundred stars , with the capacity to resolve structures forming down to mass limits of a few jupiter masses .
this system readily demonstrates the formation of small n ensembles in which the sort of behaviour described above ( binary formation , competitive accretion , ejection of low mass members ) is observed to occur .
such a one - off simulation can not be used to generate statistical results however , and so follow up simulations , tracing the evolution of large numbers of small n ensembles are required .
_ brown dwarfs and early stellar evolution _ small n - body systems still in the process of accumulating mass from an infalling envelope would observationally be seen as a class 0 or perhaps a class i source with strong outflow activity .
if brown dwarfs are formed by the disintegration of small n - body systems , it follows that the very youngest , indeed newborn , brown dwarfs should be found in the immediate vicinity of such sources . as a small triple or multiple system breaks up ,
its members start to drift out of the nascent envelope , and may on relatively short time scales ( of order a couple of thousand years ) emerge from being deeply embedded infrared sources with ample far - infrared and sub - mm emission to being optically visible t tauri like stars . in this radically different picture of early stellar evolution ,
the gradual and smooth transition between class 0 and class ii sources can be replaced by a rather abrupt transition , and the main accretion phase for the members of a multiple system is terminated not by the infalling envelope running out of gas , or outflow blowing away the last parts of the envelope , but by the newborn members `` leaving the nest '' ( reipurth 2000 ) .
considering the rapidly increasing evidence for multiplicity among the youngest stars , it is interesting to review the early idea by larson ( 1972 ) that all single low mass stars may have formed in dynamical interactions among newly born multiple systems .
one observational test of the dynamical formation model of brown dwarfs would be to study carefully the statistics of brown dwarfs in the vicinity of class 0 sources . to get a practical sense of the observed separation between a nascent brown dwarf and its siblings , assume that it is observed at a time @xmath18 [ yr ] after ejection and moving with the space velocity @xmath19 [ km s@xmath0 ] at an angle @xmath21 to the line - of - sight in a star forming region at a distance @xmath22 [ pc ] .
then the projected separation @xmath23 in arcsec is @xmath24 for a velocity of 1 km s@xmath0 , a brown dwarf moving out of a nearby ( @xmath22 @xmath25 130 pc ) cloud at an angle of 60@xmath26 to the line - of - sight will already be 5 arcmin away after 2 @xmath27 10@xmath1 yr .
also , half of all ejected brown dwarfs will move into the cloud from which they formed .
such objects will be detectable only as highly extincted and weak infrared sources .
_ brown dwarfs with stellar companions _ it has been known for some time that brown dwarfs are only rarely found as close ( less than 3 au ) companions to low mass stars ( the `` brown dwarf desert '' ) . but
recent work by gizis et al .
( 2001 ) has demonstrated that brown dwarfs are commonly companions to normal stars at large separations ( greater than 1000 au ) .
the separation distribution function of brown dwarfs in binary systems contains important information about their formation , and establishing its form more precisely will form a crucial test for any theory of brown dwarf formation .
the ejection hypothesis readily explains the currently available observations : brown dwarfs should rarely be found as close companions to stars , as in this environment they would , except for special circumstances , have continued to accrete mass at almost the same rate as their stellar companions , thus pushing through the substellar / stellar boundary at almost the same time . on the other hand ,
distant brown dwarf companions are readily expected , because not all ejections will lead to unbound systems .
a further test of the ejection scenario comes from studies of the binarity of brown dwarfs .
a number of brown dwarfs have been found to be binaries ( e.g. martn , brandner , & basri 1999 ) , but , intriguingly , they appear to be rather close binaries , whereas wide pairs ( many hundreds of au ) have so far not been found . in the ejection scenario , if two stellar embryos are ejected as a pair , they have to be relatively close in order to survive as a binary . obviously , the precise limit for survival depends not only on the mass of the pair , but also on whether the ejection occurs from a compact or a wider configuration of embryos .
numerical simulations will help to quantify this .
_ kinematics of brown dwarfs _
when a small n - body cluster dissolves , its members drift apart and blend with the other young stars in the general star forming environment , whether it is an association or a rich cluster .
the velocities of young stars are generally assumed to be similar to the turbulent velocity of the gas out of which they formed , i.e. of the order of a km per second .
if all stars are formed in small multiple systems ( larson 1972 ) , then to this we must add the mean velocity of the dispersing members .
if brown dwarfs are ejected stellar embryos , they must carry kinematic evidence reflecting their origin in small multiple systems .
to first order , the ejected member from such a system acquires a velocity comparable to the velocity attained at pericenter in the close triple encounter .
an observational test of the ejection scenario would therefore be to compare observed velocities of brown dwarfs with velocities derived in realistic numerical simulations .
however , the brown dwarfs studied must be young and belong to loose associations . in denser clusters
, two body relaxation will soon dominate the kinematics of its members , and field brown dwarfs will be dominated by objects that have evaporated from clusters and thus are similarly affected .
as discussed in section 3 , and illustrated in figure 2 , the most recent , realistic simulations show that the majority of brown dwarfs are ejected with space velocities of less than 1 km s@xmath0 , and only a small fraction have velocities larger than 2 km s@xmath0 .
one - dimensional velocities ( radial or tangential ) are correspondingly smaller .
it follows that a kinematical test of the ejection scenario will be very difficult to carry out . in the first kinematical study of brown dwarfs , joergens & guenther ( 2001 )
found that the _ radial _ velocity dispersion of nine brown dwarfs in cha i is 2.0 km s@xmath0 , whereas for t tauri stars in the same region it is 3.6 km s@xmath0 .
these numbers should be compared to the velocity dispersion of the gas , which is 1.2 km s@xmath0 in the region . as illustrated in figure 2
, there is virtually no dependence of the ejection velocity on mass , i.e. if most or all single stars have been ejected , then there should be little if any kinematic difference between brown dwarfs and surrounding young stars .
the existing , limited observations are thus consistent with our current understanding of the kinematics of ejected brown dwarfs and low mass stars .
one possible kinematic signature which may be unique to the ejection scenario is that the binaries that recoil from a dissolving small n cluster will have significantly smaller velocities than the ejected single objects .
once a young stellar association has been carefully studied to identify all binaries , this could be an interesting test .
_ signatures of youth in brown dwarfs _
the expulsion of an `` unfinished '' hydrostatic core from an infalling envelope may place limits on the amount of circumstellar material that can be brought with it .
freefloating brown dwarfs will therefore have _ finite _ reservoirs from which they can accrete and form planets .
however , it is only the outermost regions of a circumstellar disk that will be truncated , the precise values depending on the impact parameters in the close triple encounter that triggers the expulsion .
the innermost disk regions will be unaffected , and it is these regions that are responsible for the near - infrared excesses and t tauri like characteristics that are observed in very young brown dwarfs ( e.g. muench et al .
brown dwarfs are therefore expected to show mostly the same characteristics of youth as t tauri stars , and may even be able to form planets , but because of the limited amount of gas they carry with them , the period during which they display t tauri like characteristics may be more short - lived than for their heavier stellar siblings .
the discovery in recent years of numerous giant extrasolar planets orbiting other stars ( e.g. marcy , cochran & mayor 2000 ) , as well as a large number of low - luminosity objects that have been suggested to be freefloating giant planets ( e.g. lucas et al . 2001 ) suggests that a continuum may exist between such giant planets and brown dwarfs . by its definition as an object that can not attain hydrogen burning , a brown dwarf must have a mass of less than about 0.08 m@xmath28 , or 80 m@xmath29 , weakly dependent on metallicity ( chabrier & baraffe 2000 ) .
most of the extrasolar planets found around other stars have masses of order _ m_sin_i _
@xmath30 10 m@xmath29 .
the difference between the masses of giant extrasolar planets and brown dwarfs is therefore about an order of magnitude , much less than the three orders of magnitude spanned by stellar masses . with no clear dividing line between brown dwarfs and giant extrasolar planets , it is thus of considerable interest to ask which , if any , differences exist between these two classes of objects , and whether they perhaps could have a common origin .
it is indeed a possibility that freefloating giant planets are formed the same way as brown dwarfs , i.e. as ejectae from unstable multiple systems of forming stars . boss ( 2001 )
has advocated precisely such a scenario , suggesting that magnetic field tension in the collapse process has the effect of splitting the infalling material into very closely spaced fragments , from which objects as small as a jupiter - mass can be immediately ejected by dynamical interactions .
another possibility is that freefloating giant planets are formed like the giant planets in our own solar system , i.e. by later condensing out of material in a circumstellar disk .
subsequently , dynamical relaxation may ultimately kick one or more of the giant planets out of their nascent planetary system ( e.g. papaloizou & terquem 2001 ) .
obviously both mechanisms , despite their very different nature , may create freefloating planetary sized objects , which observationally do not appear to be easily distinguishable . because ejected objects lose most traces of their pre - history by their expulsion , attempts to compare brown dwarfs and giant extrasolar planets
should not be done on the basis of freefloating objects . on the other hand , bound systems , in which giant planets or brown dwarfs are orbiting a star ,
may retain critically important memories of their formation processes . and for such objects
major differences exist between the two classes of objects .
as discussed earlier , brown dwarfs are commonly found bound to stars only at very large separations .
their formation history clearly do not commonly permit them to exist closely bound to a star , as expected and readily explained by the ejection scenario .
this is in marked contrast to the increasingly well determined statistics for giant extrasolar planets orbiting other stars , which show that such objects are often located at extremely small separations , in regions that are off - limits to brown dwarfs in the ejection scenario .
if the observed separations were a smooth function of mass between the two categories of objects , one could possibly have argued that orbital evolution had led the smallest mass objects to migrate to locations close to their stars . but given that the two populations are strictly separated in the way they orbit their stars , we conclude that brown dwarfs and giant extrasolar planets , although both arguably having sizable freefloating populations , are likely to be formed in very different manners . in this interpretation ,
giant planets are formed at later times out of circumstellar disks and their growth is limited by how much disk material is available , whereas brown dwarfs are failed stars , which due to their expulsion could not take full advantage of their rich supply of inflowing gas to grow beyond the hydrogen burning limit .
|
brown dwarfs may have such low masses because they are prematurely ejected from small unstable multiple systems , while the members are still actively building up their masses .
we demonstrate that this scenario is consistent with all currently existing observations of brown dwarfs , and propose further observational tests .
we review the status of the latest realistic numerical simulations of disintegrating small n clusters , which show that many of the ejected members end up with masses that are substellar , drifting away from their birth region with velocities rarely exceeding 2 km s@xmath0 .
|
the electroweak standard model ( sm ) , together with the theory of strong interactions ( quantum chromodynamics , qcd ) , provides a comprehensive description of experimental data with remarkable consistency . in order to further test the validity of the sm predictions and restrict its only missing parameter , the higgs boson mass
, precision measurements play a key role .
the analysis of precision observables is sensitive to quantum corrections in the theoretical predictions , which depend on all the parameters of the model . in this way
, the top - quark mass , @xmath3 , had been predicted in the region where it was experimentally found .
the constraints on the higgs mass , @xmath4 , are still rather weak since @xmath4 appears only logarithmically in the leading order sm predictions .
therefore it is of high interest to further reduce experimental and theoretical uncertainties .
an important quantity in this context is the quantum correction @xmath0 in the relation of the gauge boson masses @xmath5 with the fermi constant @xmath6 and the fine structure constant @xmath7 @xcite .
this relation is established in terms of the muon decay width , which was first described in the fermi model as a four - fermion interaction @xmath8 with coupling @xmath6 .
this yields for the muon decay width @xmath9 where @xmath10 subsumes effects of the electron mass on the final - state phase space and @xmath11 denotes the qed corrections in the fermi model . including two - loop qed corrections @xcite one obtains from the measurement of the muon decay width @xcite @xmath12 .
the decay process in the sm , on the other side , involves the exchange of a w boson and additional electroweak higher order corrections . although not part of the fermi model , tree - level w propagator effects are conventionally included in eq .
( [ gfdef ] ) by means of a factor @xmath13 .
yet , this is of no numerical significance .
the relation between the fermi constant @xmath6 and the sm parameters is expressed as @xmath14 here the sm radiative corrections are included in @xmath0 ( with @xmath15 ) .
since @xmath6 is known with high accuracy it can be taken as an input parameter in equation ( [ drdef ] ) in order to obtain a prediction for the w mass ( @xmath16 is still afflicted with a considerable experimental error , @xmath17 gev @xcite ) : @xmath18 \hspace{-.5ex}. \label{mwpred}\ ] ] since @xmath0 itself depends on @xmath16 , eq .
( [ mwpred ] ) is to be understood as an implicit expression .
one expects substantial improvement on the w mass determination from future colliders , which will allow for increasingly stringent constraints on the sm from the comparison of the prediction with the experimental value for @xmath16 .
this requires an accurate theoretical determination of the quantity @xmath0 .
the one - loop result @xcite can be split into the following contributions : @xmath19 dominant corrections arise from the shift in the fine - structure constant , @xmath20 , due to large logarithms of light - fermion masses ( @xmath21 ) , and from the leading contribution @xmath22 to the @xmath23 parameter resulting from the top / bottom doublet , which enters through @xmath24 ( @xmath25 ) .
the full @xmath4dependence is contained in the remainder @xmath26 ( @xmath27 ) . furthermore , qcd corrections of @xmath28 ) @xcite and @xmath29 ) @xcite are known .
leading electroweak fermionic @xmath2 contributions were first taken into account by means of resummation relations @xcite .
different approaches for the calculation of electroweak two - loop corrections have been pursued , involving expansions for large values of @xmath4 @xcite and @xmath3 @xcite .
it turned out that both the leading @xmath30 @xcite and the next - to - leading @xmath31 @xcite coefficients in the @xmath3 expansion yield important corrections of comparable size .
therefore a complete two - loop calculation of fermionic contributions would be desirable in order to further reduce the theoretical uncertainty , in particular if one considers that these contributions are dominant already in the one - loop result .
a first step into this direction was the determination of the exact higgs mass dependence of the fermionic @xmath2 corrections to @xmath0 @xcite .
recently the full calculation of these contributions @xcite has been accomplished .
the results include all electroweak two - loop diagrams with one or two fermion loops without any expansion in @xmath3 or @xmath4 .
this talk gives an overview on the techniques and the results of this calculation .
since in the present calculation all possibly infrared divergent photonic corrections are already contained in the definition ( [ gfdef ] ) of the fermi constant @xmath6 and mass singularities are absorbed in the running of the electromagnetic coupling , @xmath16 represents the scale for the electroweak corrections in @xmath32 .
therefore it is possible to neglect all fermion masses except the top quark mass and the momenta of the external leptons so that the muon decay diagrams reduce to vacuum diagrams .
all qed contributions to the fermi model have to be excluded in the computation of @xmath0 as they are separated off in the definition of @xmath6 , see eq .
( [ gfdef ] ) .
when writing the factor @xmath33 formally as @xmath34 the decay width up to @xmath2 can be decomposed as @xmath35 & = & \gamma_{\mu,{\mbox{\scriptsize tree } } } \bigl [ 1 + 2(\delta\omega^{(\alpha ) } + \delta r^{(\alpha ) } ) + ( \delta\omega^{(\alpha ) } + \delta r^{(\alpha)})^2 + 2 \delta\omega^{(\alpha ) } \delta r^{(\alpha ) } \\[1ex ] & & \qquad\quad + 2 \delta\omega^{(\alpha^2 ) } + 2 \delta r^{(\alpha^2 ) } + { \mathcal{o}}(\alpha^3 ) \bigr ] .
\end{array}\ ] ] apart from the one - loop contributions this includes two - loop qed corrections @xmath36 and mixed contributions of qed and weak corrections @xmath37 which both thus have to be excluded in @xmath38 . for fermionic two - loop diagrams it is possible to find a one - to - one correspondence between qed graphs in fermi model and sm contributions .
the renormalization is performed in the on - shell scheme . in this context
the mass renormalization constants require the computation of two - loop self - energy diagrams with non - vanishing momentum .
all decay amplitudes and counterterm contributions have been generated with the program _
feynarts 2.2 _ @xcite .
the amplitudes are algebraically reduced by means of a general tensor integral decomposition for two - loop two - point functions with the program _ twocalc _
@xcite , leading to a fixed set of standard scalar integrals .
analytical formulae are known for the scalar one - loop @xcite and two - loop @xcite vacuum integrals whereas the two - loop self - energy diagrams can be evaluated numerically by means of one - dimensional integral representations @xcite . in order to apply an additional check
the calculations were performed within a covariant @xmath39 gauge , which introduces one gauge parameter @xmath40 , for each gauge boson .
it has been explicitly checked at the algebraic level that the gauge parameter dependence of the final result drops out .
for the determination of the one - loop counterterms ( cts ) and renormalization constants the conventions of ref .
@xcite are adopted .
two - loop renormalization constants enter via the counterterms for the transverse w propagator and the charged current vertex : @xmath41_{{\mbox{\scriptsize t } } } & = & \delta z_{(2)}^{{\mbox{\scriptsize w}}}(k^2-{m_{{\mbox{\scriptsize w}}}}^2 ) - \delta m^2_{{\mbox{\scriptsize w(2)}}}- \delta z_{(1)}^{{\mbox{\scriptsize w } } } \delta m^2_{{\mbox{\scriptsize w(1 ) } } } , \label{w - ct } \\[2ex ] \mbox{\raisebox{-8mm}{\psfig{figure = vertex.ps , width=25 mm } } } & = & i \frac{e}{\sqrt{2 } { s_{\mbox{\tiny{w } } } } } \ , \gamma_\mu \omega_- \biggl [ \delta z_{e(2 ) } - \frac{{\delta s_{\mbox{\tiny{w(2)}}}}}{{s_{\mbox{\tiny{w } } } } } + \frac{1}{2 } \left(\delta z_{(2)}^{e{\mbox{\scriptsize l}}}+\delta z_{(2)}^{{\mbox{\scriptsize w } } } + \delta z_{(2)}^{\nu { \mbox{\scriptsize l } } } \right ) \\[-1ex ] & & + \ ; \mbox{(1-loop renormalization constants ) } \biggr ] .
\nonumber\end{aligned}\ ] ] here , @xmath42 denotes the field - renormalization constant of the field @xmath43 , @xmath44 the corresponding mass ct , and @xmath45 the charge - renormalization constant .
the numbers in parentheses indicate the loop order .
the mixing angle ct , @xmath46 , is expressible through the gauge boson mass cts . throughout this paper
, the two - loop contributions always include the subloop renormalization .
the on - shell masses are defined as the position of the propagator poles .
starting at the two - loop level , it has to be taken into account that there is a difference between the definition of the mass @xmath47 as the real pole of the propagator @xmath48 , @xmath49 and the real part @xmath50 of the complex pole , @xmath51 the imaginary part of the complex pole is associated with the width @xmath52 .
the definition ( [ complex ] ) yields for the w mass ct @xmath53 whereas for the real pole definition the last term of eq .
( [ mwct ] ) is missing .
@xmath54 denote the transverse w self - energy and its momentum derivative .
similar expressions hold for the z boson .
the w and z mass cts determine the two - loop mixing angle ct , @xmath46 , which has to be gauge invariant since @xmath55 is an observable quantity . with the use of a general @xmath39 gauge
it could be explicitly checked that @xmath46 is gauge - parameter independent for the complex - pole mass definition , whereas the real - pole definition leads to a gauge dependent @xmath46 .
this is in accordance with the expectation from s - matrix theory @xcite , where the complex pole represents a gauge - invariant mass definition .
it should be noted that the mass definition via the complex pole corresponds to a breit - wigner parameterization of the resonance shape with a constant width .
for the experimental determination of the gauge boson masses , however , a breit - wigner ansatz with a running width is used .
this has to be accounted for by a shift of the values for the complex pole masses @xcite , @xmath56 which yields the relations @xmath57 for @xmath58 and @xmath59 the experimental numbers are taken . the w mass is a calculated quantity , and therefore also a theoretical value for the w boson width should be applied here .
the results above are obtained from the approximate , but sufficiently accurate expression for the w width , @xmath60 at the subloop level , also the faddeev - popov ghost sector has to be renormalized .
the gauge - fixing sector for the gauge fields @xmath61 , @xmath62 , @xmath63 ( @xmath64 , @xmath65 denote the unphysical higgs scalars ) @xmath66 @xmath67 @xmath68 @xmath69 does not need renormalization .
accordingly , one can either introduce the gauge - fixing term after renormalization or renormalize the gauge parameters in such a way that they compensate the renormalization of the fields and masses .
both methods ensure that no counterterms arise from the gauge - fixing sector but they differ in the treatment of the ghost lagrangian , which is given by the variation of the functionals @xmath70 under infinitesimal gauge transformations @xmath71 , @xmath72 in the latter case , which was applied in this work , additional counterterm contributions for the ghost sector arise from the gauge parameter renormalization .
the parameters @xmath73 in ( [ gaugefix ] ) are renormalized such that their cts @xmath74 exactly cancel the contributions from the renormalization of the fields and masses and that the renormalized gauge parameters comply with the @xmath39 gauge .
in four dimensions the algebra of the @xmath1matrix is defined by the two relations @xmath75 \mbox{tr } \bigl\ { \gamma_5 \gamma^\mu\gamma^\nu\gamma^\rho\gamma^\sigma \bigr\ } = 4i \epsilon^{\mu\nu\rho\sigma}\mbox{. } \label{trg4}\end{aligned}\ ] ] it is impossible to translate both relations simultaneously into @xmath76 dimensions without encountering inconsistencies @xcite . a certain treatment of @xmath1 might break symmetries , i.e. violate slavnov - taylor ( st ) identities which would have to be restored with extra counterterms . even after this procedure a residual scheme dependence can persist which is associated with @xmath77-tensor expressions originating from the treatment of ( [ trg4 ] ) .
such expressions can not be canceled by counterterms .
if they broke st identities this would give rise to anomalies .
t hooft and veltman
@xcite suggested a consistent scheme which was formulated by breitenlohner and maison @xcite as a separation of the first four and the remaining dimensions of the @xmath78-matrices ( hvbm - scheme ) .
it has been shown @xcite that the sm with hvbm regularization is anomaly - free and renormalizable .
this shows that @xmath77-tensor terms do not get merged with divergences .
the naively anti - commuting scheme , which is widely used for one - loop calculations , extends the rule ( [ anticomm ] ) to @xmath48 dimensions but abandons ( [ trg4 ] ) , @xmath79 \mbox{tr } \bigl\ { \gamma_5 \gamma^\mu\gamma^\nu\gamma^\rho\gamma^\sigma \bigr\ } = 0.\end{aligned}\ ] ] this scheme is unambiguous but does not reproduce the four - dimensional case . in the sm particularly triangle diagrams ( fig .
[ fig : tri ] ) containing chiral couplings are sensitive to the @xmath1problem . for the presented work one - loop triangle diagrams
have been explicitly calculated in both schemes . while the naive scheme immediately respects all st identities the hvbm scheme requires the introduction of additional finite counterterms . even after this procedure
finite differences remain between the results of the two schemes , showing that the naive scheme is inapplicable in this case . in the calculation of @xmath0 triangle diagrams appear as subloops of two - loop charged current ( cc ) vertex diagrams ( fig .
[ fig : ccvert ] ) .
one finds that for the difference terms between both schemes this loop can be evaluated in four dimensions without further difficulties .
this can be explained by the fact that renormalizability forbids divergent contributions to @xmath77-tensor terms from higher loops in the hvbm scheme .
the @xmath77-tensor contributions from the triangle subgraph in the hvbm scheme meet a second @xmath77-tensor term from the outer fermion lines in fig .
[ fig : ccvert ] , thereby resulting in a non - zero contribution to @xmath0 . computations in the hvbm scheme can get very tedious because of the necessity of additional counterterms .
therefore another method shall be examined .
one can consider a `` mixed '' scheme that uses both relations ( [ anticomm ] ) and ( [ trg4 ] ) in @xmath48 dimensions despite their mathematical inconsistency .
this scheme is plagued by ambiguities of @xmath80 .
when applied to the calculation of one - loop triangle diagrams the results immediately respect all st identities and differ from the hvbm results only by terms of @xmath80 , @xmath81 since for the difference terms the second loop can be evaluated in four dimensions , this also holds for the two - loop cc diagrams , @xmath82 thus the mixed scheme , despite being mathematically inconsistent , can serve as a technically easy prescription for the correct calculation of the cc two - loop contributions .
in the previous sections the characteristics of the calculation of electroweak two - loop contributions to @xmath0 have been pointed out .
combining the fermionic @xmath2 contributions with the one - loop and the qcd corrections yields the total result @xmath83 here @xmath84 symbolize one and two fermionic loops respectively .
[ fig : deltar ] shows that both the qcd and electroweak two - loop corrections give sizeable contributions of 1015% with respect to the one - loop result . in fig .
[ fig : mwpred ] the prediction for @xmath16 derived from the result ( [ drtotal ] ) and the relation ( [ mwpred ] ) is compared with the experimental value for @xmath16 .
dotted lines indicate one standard deviation bounds .
the main uncertainties of the prediction originate from the experimental errors of @xmath85 gev @xcite and @xmath86 @xcite .
it can be noted that light higgs masses are favored by this analysis .
further implications of the precision calculation of @xmath16 are discussed in @xcite .
these results can be compared with the results obtained by expansion of the two - loop contributions up to next - to - leading order in @xmath3 @xcite .
the predicted values for @xmath16 for several values of @xmath4 are given in tab .
[ tab : mwcomp ] . ? = ?
@r@r@ r@r@r ' '' '' @xmath4 & @xmath87 & @xmath88 & @xmath89 + ' '' '' [ gev ] & [ gev ] & [ gev ] & [ mev ] + @xmath90 & @xmath91 & @xmath92 & @xmath93 + @xmath94 & @xmath95 & @xmath96 & @xmath97 + @xmath98 & @xmath99 & @xmath100 & @xmath101 + @xmath102 & @xmath103 & @xmath103 & @xmath104 + @xmath105 & @xmath106 & @xmath107 & @xmath108 + agreement is found between the results with maximal deviations of about 4 mev in @xmath16 .
the theoretical uncertainty due to missing higher order contributions can be estimated as follows .
the missing @xmath2 purely bosonic corrections can be judged by means of resummation relations to be very small ( @xmath109 mev effect on @xmath16 for a light higgs ) .
an estimate of the @xmath110 terms can be obtained from the renormalization scheme dependence of the two - loop result yielding about 23 mev , and the missing higher order qcd corrections were estimated to be about 45 mev @xcite . adding this up linearly , one arrives at a total uncertainty of about 7 mev for the @xmath16prediction at low higgs masses ( @xmath111 gev ) .
in this talk the realization of an exact two - loop calculation of fermionic contributions in the full electroweak sm and its application to the precise computation of @xmath0 were presented .
numerical results and an estimate of the remaining uncertainties were given and might serve as ingredient for future sm fits . 1 cm a. sirlin , phys .
d22 ( 1980 ) 971 . t. van ritbergen , r. stuart , phys . lett . b437 ( 1998 ) 201 ; phys .
82 ( 1999 ) 488 .
m. steinhauser , t. seidensticker , phys .
lett . b467 ( 1999 ) 271 . c. caso et al .
c3 ( 1998 ) 1 and 1999 off - year partial update ( http://pdg.lbl.gov/ ) .
lep electroweak working group , winter 99/00 results + ( http://lepewwg.web.cern.ch/lepewwg/ ) .
marciano , a. sirlin , phys .
d22 ( 1980 ) 2695 .
a. djouadi , nuovo cim . 100a ( 1988 ) 357 ; + b.a .
kniehl , nucl .
b347 ( 1990 ) 86 ; + f. halzen , b.a .
kniehl , nucl .
b353 ( 1991 ) 567
. l. avdeev , j. fleischer , s.m .
mikhailov , o. tarasov , phys .
b336 ( 1994 ) 560 ; err .
: b349 ( 1995 ) 597 ; + k. chetyrkin , j. khn , m. steinhauser , phys .
lett . b351 ( 1995 ) 331 ; phys .
lett . 75 ( 1995 ) 3394 .
a. sirlin , phys .
d29 ( 1984 ) 89 ; + m. consoli , w. hollik , f. jegerlehner , phys .
227 ( 1989 ) 167 .
j. van der bij , m. veltman , nucl .
b231 ( 1984 ) 205 .
j. van der bij , f. hoogeveen , nucl .
b283 ( 1987 ) 477 .
r. barbieri , m. beccaria , p. ciafaloni , g. curci , a. vicere , phys .
b288 ( 1992 ) 95 , err .
: b312 ( 1993 ) 511 ; nucl .
b409 ( 1993 ) 105 ; + j. fleischer , o.v .
tarasov , f. jegerlehner , phys .
lett . b319 ( 1993 ) 249 ; phys .
d51 ( 1995 ) 3820 ; + g. degrassi , s. franchiotti , p. gambino , int . j. mod
a10 ( 1995 ) 1337 .
g. degrassi , p. gambino , a. vicini , phys .
b383 ( 1996 ) 219 .
g. degrassi , p. gambino , a. sirlin , phys .
b394 ( 1997 ) 188 .
s. bauberger , g. weiglein , phys .
lett . b419
( 1998 ) 333 .
a. freitas , w. hollik , w. walter , g. weiglein , hep - ph/0007091 ; + see also a. freitas , diplomarbeit ( universitt karlsruhe , 1999 , ( http://www-itp.physik.uni-karlsruhe.de/prep/diploma/ ) .
j. kblbeck , m. bhm , a. denner , comp .
. commun . 60
( 1990 ) 165 ; + h. eck , j. kblbeck , guide to feynarts 1.0 ( universitt wrzburg , 1992 ) ; + t. hahn , feynarts 2.2 user s guide ( universitt karlsruhe , 1998 , http://www.feynarts.de// ) . g. weiglein , r. scharf , m. bhm , nucl . phys . b416 ( 1994 ) 606 ; + g. weiglein , r. mertig , r. scharf , m. bhm in new computing techniques in physics research 2 , ed .
j. perret - gallix ( world scientific , singapore , 1992 ) 617 .
g. t hooft , m. veltman , nucl .
b153 ( 1979 ) 365 .
a. i. davydychev , j. b. tausk , nucl .
b397 ( 1993 ) 123 .
s. bauberger , f. a. berends , m. bhm , m. buza , g. weiglein , nucl .
b37 ( 1994 ) 95 ; + s. bauberger , f. a. berends , m. bhm , m. buza , nucl .
b434 ( 1995 ) 383 ; + s. bauberger , m. bhm , nucl .
b445 ( 1995 ) 25 .
a. denner , fortschr .
41 ( 1993 ) 307 .
a. sirlin , phys .
67 ( 1991 ) 2127 ; phys .
b267 ( 1991 ) 240 ; + r. g. stuart , phys . lett .
b262 ( 1991 ) 113 ; phys .
70 ( 1993 ) 3193 ; + m. passera , a. sirlin , phys .
d58 ( 1998 ) 113010 ; + p.
gambino , p.a .
grassi , hep - ph/9907254 .
d. bardin , a. leike , t. riemann , m. sachwitz , phys .
b206 ( 1988 ) 539 .
g. t hooft , m. veltman , nucl .
b44 ( 1972 ) 189 .
p. breitenlohner , d. maison , comm .
52 ( 1977 ) 11 .
d. j. gross , r. jackiw , phys .
d6 ( 1972 ) 477 ; + g. costa , t. marinucci , m. tonin , j. julve , nuovo cim . 38a ( 1977 ) 373 .
s. eidelman , f. jegerlehner , z. phys .
c67 ( 1995 ) 585 ; + m. steinhauser , phys .
lett . b429
( 1998 ) 158 .
b. n. taylor , eur .
c3 ( 1998 ) 69 .
g. weiglein , these proceedings .
a. sirlin , phys .
b348 ( 1995 ) 201 .
p. gambino , proceedings of radcor 98 , barcelona , ed .
j. sol ( world scientific , singapore , 1999 ) , 456 - 470 ( hep - ph/9812332 ) ; + g. degrassi , acta phys .
b29 ( 1998 ) 2683 .
|
the computation of the correction @xmath0 in the w - z mass correlation , derived from muon decay , is described at the two - loop level in the standard model .
technical aspects which become relevant at this level are studied , e.g. gauge - parameter independent mass renormalization , ghost - sector renormalization and the treatment of @xmath1 .
exact results for @xmath0 and the w mass prediction including @xmath2 corrections with fermion loops are presented and compared with previous results of a next - to - leading order expansion in the top - quark mass .
|
we conduct a series of experiments with a special interest on a penetration process and instabilities arisen on a liquid jet impinged to a liquid of the same kind flowing in a channel .
the impinged jet penetrates into the flowing bath accompanying with entrainment of the ambient immiscible gas , which results in the impinged jet wrapped by the entrained gas as a sheath. this sheath formation enables the impinged jet to survive in the fluid in the channel without coalescing until the entrained - air sheath breaks down .
occasionally a cap of the entrained air is formed at the tip of the penetrated jet , and the jet elongates like a long balloon .
|
dynamic behaviors of the penetrated jet and the departure of the bubble of wrapping gas at the tip of the collapsing jet observed by use of a high - speed camera are included in this fluid dynamics video .
|
one of the most important prediction of einstein theory of relativity is gravitational radiation . since the very precise measurement of the orbital decay in the binary pulsar b1913 + 16 system by hulse and taylor ,
the existence of gravitational waves ( gw ) has been indirectly proved and general relativity has passed another quite constraining test . due to
the emission of gw , binary neutron stars ( ns ) decrease their orbital separation and finally merge .
the evolution of a binary system can be separated into three phases : point - like inspiral where orbital separation is much larger than the ns radius , hydrodynamical inspiral where orbital separation is just a few times larger than the radius of the ns so that hydrodynamics play an important role , and merger in which the two stars coalesce dynamically .
the gw signal of the terminal phases ( the hydrodynamical phase or the merger phase ) of inspiraling binary can bring the information about the stellar structure .
in particular it may be possible to impose constraints on the equation of state of ns .
it is still an open question whether the core of ns consists mainly of superfluid neutrons or an exotic matter like kaon condensations , pion condensations or strange quark matter .
as suggested by bodmer ( 1971 ) the absolute true ground state of nuclear matter may be a state of deconfined up , down and strange quarks ( since energy per baryon for strange matter is lower than the energy per baryon for fe@xmath0 ) .
if it is true then objects made of such matter so called strange stars ( ss ) could exists ( witten 1984 ) .
ss are currently considered as a possible alternative to ns as compact objects ( see e.g. madsen 1999 for a review and gondek - rosiska et al .
2003 ) . up to now
, majority of the relativistic calculations of the terminal phase of inspiral have been done for binary systems containing ns described by a simplified equation of state of dense matter so called polytropic eos ( taniguchi & gourgoulhon 2003 ) . in the paper we present the results of our studies on the hydrodynamical phase of inspiraling binary systems containing equal mass strange stars ( according to the recent population synthesis calculations ( bulik , gondek - rosiska and belczyski , 2004 ) a significant fraction of the observed binary ns in gw will contain stars with masses @xmath1 ) .
we compare the evolution of ss - ss systems with ns - ns systems in order to find any characteristic features in the gw waveform that will help to distinguish between ss and ns .
as already mentioned , ss are composed of deconfined up , down and strange quarks .
typically , they are modeled with an equation of state based on the mit - bag model in which quark confinement is described by an energy term proportional to the volume ( fahri et al . , 1984 ) .
ss are self - bound objects , having high density ( @xmath2 ) at the surface .
there are three physical quantities entering the mit - bag model : the mass of the strange quarks , @xmath3 , the bag constant , @xmath4 , and the strength of the qcd coupling constant @xmath5 . in the numerical calculations reported in the present paper we consider three different mit - bag models ( corresponding to three different sets of the model parameters ) : * sqs0 * - the standard mit bag model : @xmath6 , @xmath7 , @xmath8 ; * sqs1 * - the simplified mit bag model with @xmath9 , @xmath10 ; @xmath11 ; * sqs2 * - the extreme mit bag model ( relatively low strange quark mass and @xmath4 but high @xmath5 ) : @xmath12 , @xmath13 , @xmath14
in the hydrodynamical phase , since the timescale of orbital shrinking due to the emission of gw is longer than the orbital period , one may consider a binary ns system to be in quasiequilibrium state .
for given eos , we construct so called _ evolutionary sequence _ by calculating a sequence of quasiequilibrium configurations with constant baryon mass for decreasing orbital separation . the second assumption is to consider a conformally flat metric , which corresponds to the _ isenberg - wilson - mathews _ approximation of general relativity . in this approximation ,
the spacetime metric takes the form : ds^2 = -(n^2 -b_i b^i ) dt^2 -2b_i dt dx^i + a^2 f_ij dx^i dx^j , [ eq : metric ] where @xmath15 is the lapse , @xmath16 the conformal factor , @xmath17 the shift vector and @xmath18 the flat spatial metric .
another assumption concerns the fluid motion inside each star , here we considered irrotational binaries .
) versus frequency of gw along two irrotational equilibrium sequences .
solid and long - dashed lines correspond to strange quark stars described by mit bag model and ns described by polytropic eos respectively .
a diamond indicates the marginally stable orbit .
a dotted line corresponds to 3rd post - newtonian calculations for point masses .
, width=264 ] in order to calculate the last orbits of inspiral phase of binary ns and ss we use highly accurate numerical code which solves the five elliptic equations for the gravitational field ( for @xmath16 , @xmath15 and @xmath17 ) , supplemented by an elliptic equation for the velocity potential in the case of irrotational flows ( see limousin , gondek - rosiska & gourgoulhon 2004 for boundary conditions in the case of ss ) . in fig .
1 we show the evolution of equal mass binary ns ( a long - dashed line ) and ss described by sqs1 model ( a solid line ) having total gravitational mass @xmath19 at infinity . the binding energy is defined as the difference between @xmath20 ( see taniguchi & gourgoulhon 2003 ) and the total mass of the system at infinity . we see a good agreement with 3pn calculations for big distances i.e. small frequencies , where the internal structure of the stars is not the predominant effect . in order to compare results for ss and ns we take a polytrope giving the same gravitational mass and the same compaction parameter @xmath21 at infinity as obtained for static ss .
the minimum of energy ( shown as a diamond ) for the evolutionary sequence of ss corresponds to the appearance of a dynamical instability for binaries ( the innermost stable circular orbit ( isco ) ) .
the frequency of the isco is a potentially observable parameter by the gw detectors .
we do nt see the isco for ns .
the sequence of ns terminates by the mass - shedding limit ( corresponding to exchange of matter between two stars ) .
different evolutions of ns and ss stem from the fact that ss are principally bound by an additional force , strong interaction between quarks ( for the same distances there are less deformed than ns ) . in fig 2 .
we present the evolution of binary ss , with total mass 2.7 @xmath22 , described by three different sets of eos parameters of the mit bag model .
we see that the frequency of gw at the isco strongly depends on the compaction parameter - the higher it is , the higher the frequency at the isco .
detection of gw may help to impose constraints on the eos of ns and ss and to find the ground state of matter at high densities .
partially supported by the kbn grants 5p03d.017.21 and pbz - kbn-054/p03/2001 ; by the `` bourses de recherche 2004 de la ville de paris '' and by the associated european laboratory astro - pf ( astrophysics poland - france ) .
bodmer , 1971 , phys .
d , 4 , 1601 bulik , t .
, gondek - rosiska , d. , belczyski , k. , 2004 , mnras , 352 , 1372 fahri , e. , & jaffe , r.l . , 1984 , phys .
d , 30 , 2379 gondek - rosiska , d . ,
gourgoulhon , e . , haensel , p. , 2003 ,
a&a , 412 , 777 limousin , f .
, gondek - rosinska , d . ,
gourgoulhon , e . , 2004 , in preparation madsen , 1999 , lect .
notes in phys .
|
inspiraling compact binaries are expected to be the strongest sources of gravitational waves for virgo , ligo and other laser interferometers .
we present the first computations of quasi - equilibrium sequences of compact binaries containing two strange quark stars ( which are currently considered as a possible alternative to neutron stars ) .
we study a precoalescing stage in the conformal flatness approximation of general relativity using a multidomain spectral method .
a hydrodynamical treatment is performed under the assumption that the flow is irrotational .
|
for an eigenfunction @xmath0 corresponding to the @xmath1-th eigenvalue @xmath2 ( counted with multiplicity ) of the laplace operator in a bounded regular domain @xmath3 , we denote by @xmath4 the number of nodal domains of @xmath0 .
a famous result by courant ( see @xcite ) states that @xmath5 .
if @xmath6 , then we say that the eigenpair @xmath7 ( or just the eigenvalue @xmath2 ) is courant sharp .
it is proved in @xcite that , for general planar domains , and with dirichlet or neumann boundary conditions , the courant sharp situation occurs for a finite number of eigenvalues only .
note that in the case of neumann the additional assumption that the boundary is piecewise analytic should be imposed due the use of a theorem by toth
zelditch @xcite counting the number of nodal domains whose closure is touching the boundary .
in the recent years , the question of determining the courant sharp cases reappears in connection with the determination of minimal spectral partitions in the work of helffer hoffmann - ostenhof
terracini @xcite .
the courant sharp situation was analyzed there in the case of the irrational rectangle and in the case of the disk for dirichlet boundary condition .
the case of anisotropic ( irrational ) tori is solved in @xcite .
recently the courant sharp cases were identified in the cases of @xmath3 being a square with dirichlet boundary conditions imposed @xcite , and @xmath3 being the two - sphere @xcite . here , our aim is to do the same detailed analysis in the case of @xmath3 being a square with neumann boundary conditions .
we let @xmath8 and denote by @xmath9 the self - adjoint neumann laplacian in @xmath10 .
this operator has eigenvalues @xmath11 generated by the set @xmath12 .
a basis for the eigenspace corresponding to the eigenvalue @xmath13 is given by @xmath14 .
pleijel was in particular referring to figures appearing in the book of pockels @xcite ( and partially reproduced in @xcite ) like in the figure [ pockel ] . in the book of pockels ( 1891).,title="fig:",width=377 ] [ pockel ] [ thm : main ]
there exists a courant sharp eigenpair @xmath7 if and only if @xmath15 .
the courant sharpness of eigenvalues @xmath16 , @xmath17 and @xmath18 follows from lemma [ lem : p0 ] and the courant sharpness of @xmath19 and @xmath20 follows from lemma [ lem : pp ] .
these cases are illustrated in figure [ fig : cs ] .
they correspond to the zero sets of the following eigenfunctions : * @xmath21 : @xmath22 ; * @xmath23 : @xmath24 ( with @xmath25 in figure [ fig : cs ] ) ; * @xmath26 : @xmath27 ; * @xmath28 : @xmath29 ; * @xmath30 : @xmath31 . , @xmath23 , @xmath26 , @xmath28 , @xmath30 . in each example
the black and white areas represent nodal domains where the function has different sign.,title="fig:",width=75 ] 0.5 cm , @xmath23 , @xmath26 , @xmath28 , @xmath30 . in each example
the black and white areas represent nodal domains where the function has different sign.,title="fig:",width=75 ] 0.5 cm , @xmath23 , @xmath26 , @xmath28 , @xmath30 . in each example
the black and white areas represent nodal domains where the function has different sign.,title="fig:",width=75 ] 0.5 cm , @xmath23 , @xmath26 , @xmath28 , @xmath30 . in each example
the black and white areas represent nodal domains where the function has different sign.,title="fig:",width=75 ] 0.5 cm , @xmath23 , @xmath26 , @xmath28 , @xmath30 . in each example
the black and white areas represent nodal domains where the function has different sign.,title="fig:",width=75 ] the proof of theorem [ thm : main ] is divided into several lemmas and propositions . although following the general scheme proposed by .
pleijel @xcite and completed in @xcite for the dirichlet case , the realization of the program in the case of neumann is more difficult and finally involves a combination of arguments present in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and @xcite .
first we reduce to a finite number of possible courant sharp cases in section [ section2 ] . in section [ section3 ]
we use different symmetry arguments . in section [ section4 ] , we consider two families of eigenfunctions corresponding to @xmath32 and @xmath33 for which a complete description is easy .
section [ section5 ] gives the general approach for the analysis of the critical points and the boundary points together with a rough localization of the zero set initiated by a. stern : the chessboard localization .
the rest of the cases , which needs a separate treatment , are taken care of in sections [ section6 ] and [ section7 ] . in section [ section8 ]
we indicate how one can improve the estimates , if striving for optimal results .
finally , in section [ sec : table ] we give a list of all eigenvalues together with a reference to the lemma in which they are treated .
we conclude by a short discussion on open problems .
proposition [ prop : red1 ] below reduces our study to a finite number of eigenvalues .
we provide animations showing the nodal domains in all finite cases studied where the eigenspace is two - dimensional .
given an eigenfunction @xmath34 , we introduce the subset @xmath35 as the union of nodal domains of @xmath34 that do not touch the boundary of @xmath3 , except at isolated points .
we also introduce @xmath36 as the union of nodal domains of @xmath34 not belonging to @xmath37 .
we also denote by @xmath38 and @xmath39 the number of nodal domains of @xmath34 restricted to @xmath37 and @xmath40 , respectively .
it is clear that @xmath41 from @xcite we know that if @xmath7 is an eigenpair of @xmath9 then @xmath42 moreover , we can write @xmath43 as a finite union of pairwise disjoint nodal domains for @xmath0 . the faber krahn inequality @xcite for each inner nodal domain @xmath44 says @xmath45 where @xmath46 denotes the area of @xmath47 and @xmath48 the first positive zero of the bessel function @xmath49 . summing , we get @xmath50 [ prop : red1 ] assume that @xmath7 is a courant sharp eigenpair
. then @xmath51 .
let @xmath52 denote the number of eigenvalues strictly less than @xmath53 , counting multiplicity .
the weyl law @xcite says that @xmath54 but we need the following universal lower bound for the neumann problem in the square obtained by direct counting ( see @xcite for the dirichlet case with the correction mentioned in @xcite ) : @xmath55 assume that @xmath7 is a courant sharp eigenpair .
the theorem of courant implies that @xmath56 and @xmath57 .
inserting this into gives us @xmath58 combining this with and , and the estimate @xmath59 , @xmath60 this inequality is false if @xmath61 .
depending on the cases , we can consider many variants of the intermediate steps in the proof of proposition [ prop : red1 ] and introduce small useful improvements which can be used directly .
[ lemma2.2 ] assume that @xmath7 is an eigenpair of @xmath9 .
then @xmath62 this follows immediately from and together with the fact that @xmath63 must be an integer . for @xmath64 ,
we denote by @xmath65 .
[ lemma2.3 ] assume that @xmath7 is an eigenpair of @xmath9 .
then @xmath66 we observe that @xmath67 .
hence , @xmath68 [ cor2.3 ] the eigenvalues @xmath2 , where @xmath1 is one of 86 , 9596 , 99100 , 103104 , 113 , 118119 , 120121 , 128142 , 147208 , are not courant sharp .
assume that @xmath1 is such that @xmath69 .
then , a numerical calculation shows that @xmath70 for the @xmath1 mentioned in the statement .
symmetry arguments will play an important role in the analysis of the courant sharp situation .
these ideas appear already in the case of the harmonic oscillator and the sphere in contributions by j. leydold @xcite .
we introduce the notation @xmath71 we will often write just @xmath72 or @xmath73 . for eigenvalues of @xmath9 of multiplicity two , the family @xmath74 , @xmath75 will give all possible eigenfunctions ( up to multiplication by a non zero constant ) .
moreover , the basis of our arguments are the rich symmetry of the trigonometric functions .
the role of the antipodal map in the case of the sphere is now replaced in the case of the sphere by the map : @xmath76 a finer analysis will involve the finite group generated by the maps @xmath77 and @xmath78 .
we introduce @xmath79 , the neumann laplacian restricted to the antisymmetric space @xmath80 the spectrum of this laplacian is given by @xmath81 with @xmath82 odd .
we denote by @xmath83 the sequence of eigenvalues of @xmath79 , counted with multiplicity .
then each odd @xmath2 equals @xmath84 for some @xmath85 .
the next lemma is an adaptation of courant s theorem in this subspace .
[ lem : antisymmetric ] assume that @xmath7 is an eigenpair of @xmath9 , with @xmath2 odd , and let @xmath85 be such that @xmath86 .
then @xmath4 is even , and @xmath87 the proof is inspired by a proof of leydold @xcite ( used in the case of the sphere ) .
see also leydold @xcite and brard - helffer @xcite for the case of the harmonic oscillator . by assumption
we have @xmath88 this implies that @xmath4 is even and that the family of nodal domains is the disjoint union of @xmath89 pairs , each pair consisting of two disjoint open sets exchanged by @xmath90 . restricting @xmath0 to each pair
, we obtain an @xmath89-dimensional antisymmetric space whose energy is bounded by @xmath2 .
hence @xmath91 by the min - max principle and @xmath92 .
thus @xmath93 .
[ cor : antisymmetric ] the eigenvalues @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 , @xmath110 , @xmath111 , @xmath112 , @xmath113 , @xmath114 , @xmath115 , @xmath116 , @xmath117 , @xmath118 and @xmath119 are not courant sharp .
we let @xmath120 denote the neumann laplacian restricted to the symmetric space @xmath121 the spectrum of this laplacian is given by @xmath81 with @xmath82 even .
we denote by @xmath122 the sequence of eigenvalues of @xmath120 , counted with multiplicity .
each even @xmath2 equals @xmath123 for some @xmath85 .
the next lemma is an adaptation of courant s theorem in this subspace .
[ lem : symmetric ] let @xmath7 be an eigenpair of @xmath9 , with even @xmath2 , and let @xmath85 be such that @xmath124 .
then @xmath87 it is again inspired by a proof of leydold @xcite ( used in the case of the sphere ) . by assumption
we have @xmath125 .
this implies that the family of nodal domains is the disjoint union of @xmath89 pairs , each pair consisting of two disjoint open sets exchanged by @xmath90 and of @xmath126 symmetric open sets .
hence we have : @xmath127 restricting @xmath0 to each pair , or to each symmetric open set , we obtain an @xmath128-dimensional antisymmetric space whose energy is bounded by @xmath2 .
hence @xmath129 by the min - max principle and @xmath130 .
thus @xmath131 .
[ cor : symmetric ] the eigenvalues @xmath132 , @xmath133 , @xmath134 , @xmath135 , @xmath136 , @xmath137 , @xmath138 , @xmath139 , @xmath140 , @xmath141 , @xmath142 , @xmath143 , @xmath144 , @xmath145 , @xmath146 , @xmath147 , @xmath148 , @xmath149 , and @xmath150 are not courant sharp .
next , we let @xmath151 denote the neumann laplacian restricted to the anti - symmetric space @xmath152 the spectrum of this laplacian is given by @xmath81 with @xmath153 and @xmath154 odd .
we denote by @xmath155 the sequence of eigenvalues of @xmath151 , counted with multiplicity .
the next lemma is an adaptation of courant s theorem in this subspace .
[ lem : antimirror ] assume that @xmath7 is an eigenpair of @xmath9 , with @xmath2 even and @xmath156 .
then @xmath157 for @xmath85 such that @xmath158 .
moreover , @xmath4 is divisible by @xmath159 . by assumption @xmath160 and @xmath161 .
this implies that the nodal domains is the disjoint union of @xmath89 quadruples .
hence we have : @xmath162 this proves the last statement . restricting @xmath0 to each quadruple
, we obtain an @xmath89-dimensional space whose energy is bounded by @xmath2 .
hence @xmath163 by the min - max principle and @xmath164 .
thus @xmath165 .
if , for all pairs @xmath166 of non - negative integers such that @xmath167 , it holds that @xmath153 and @xmath154 are odd , then there exists an @xmath85 such that @xmath168 .
[ cor : antimirror ] the eigenvalues @xmath169 , @xmath170 , @xmath132 , @xmath171 , @xmath133 , @xmath172 , @xmath173 , @xmath174 , @xmath175 , @xmath136 , @xmath176 , @xmath138 , @xmath140 , @xmath177 , @xmath178 , @xmath142 , @xmath179 , @xmath143 , @xmath180 , @xmath181 , @xmath146 , @xmath148 and @xmath149 are not courant sharp .
[ lem : pandqeven ] assume that @xmath182 has at least @xmath183 solutions for @xmath184 ( @xmath185 ) and that @xmath186 has at least @xmath187 solutions ( @xmath188 ) for @xmath189 .
then @xmath190 if , moreover , @xmath191 , @xmath192 the function @xmath193 is even in the lines @xmath194 and @xmath195 .
we note that for each zero described in the statement ( except the biggest one ) , we count for @xmath196 one nodal domain two times .
the one in the middle is subtracted three times if @xmath197 and four times if @xmath198 .
[ cor : pandqeven ] the eigenvalues @xmath199 and @xmath200 are not courant sharp . [ lemma3.8 ] for odd eigenvalues of multiplicity two , to get the maximum number of possible nodal domains , it is sufficient to study @xmath74 for @xmath202 .
as we have seen the odd eigenvalues correspond to the case @xmath82 odd .
assume , without loss of generality , that @xmath153 is even and @xmath154 is odd .
then the statement follows directly from the relations @xmath203 note that holds for all @xmath153 and @xmath154 , not only for @xmath82 odd .
in this case we are able to calculate exactly the maximum number of nodal domains .
we start with the first non - trivial case .
[ lem:10 ] let @xmath206 be an eigenfunction corresponding to @xmath207
. then @xmath208 .
moreover , the nodal line will either go from the side @xmath209 to the side @xmath210 or from the side @xmath211 to the side @xmath212 ( or be a diagonal ) . in any case
it will not be a loop .
since @xmath213 is the second eigenvalue , it follows directly that @xmath208 .
the eigenfunction @xmath206 will have the form @xmath214 if @xmath215 then @xmath216 if @xmath217 or @xmath218 , then the equations and has exactly one solution each , and the equations and has no solutions . if @xmath219 or @xmath220 , then the opposite situation holds . in the remaining cases
the nodal lines are just straight lines . if @xmath221 then the nodal line is just @xmath194 . if @xmath222 then it is @xmath223 . if @xmath224 then it is @xmath195 and if @xmath225 then it is @xmath226 .
[ lem : p0 ] assume that @xmath227 is an eigenpair of @xmath9 of multiplicity two , corresponding to @xmath204 and @xmath228 .
then @xmath229 moreover , in each situation , equality holds for some function @xmath34 in the eigenspace .
the case @xmath230 is clear since then the eigenfunction is just a constant , having one nodal domain .
the case @xmath231 ( and @xmath232 ) was taken care of in lemma [ lem:10 ] . for @xmath204 , @xmath233 ,
the eigenfunctions looks like @xmath234 note that , for all @xmath201 , the function @xmath235 , @xmath189 , @xmath184 is exactly the function in the eigenspace corresponding to the case @xmath231 , whose nodal domains we know of from lemma [ lem:10 ] .
the function @xmath236 is reconstructed by taking its values in the square @xmath237 , @xmath238 , and then `` folding '' it evenly over the whole square .
indeed , for integers @xmath183 , @xmath239 if @xmath240 then the @xmath241 has only one nodal line , going from one side to its opposite side .
when folding , this results in exactly @xmath242 nodal domains .
see figure [ fig:3 - 0 - 1 ] for a typical case . if @xmath222 , then , in the square @xmath243 , @xmath238 , @xmath241 has one nodal line , @xmath244 . folding this square gives indeed @xmath245 nodal domains if @xmath153 is odd , and @xmath246 nodal domains if @xmath153 is even .
this is illustrated ( as the left pictures ) in figure [ fig:3 - 0-special ] and [ fig:4 - 0-special ] .
if @xmath225 then , in the square @xmath243 , @xmath247 @xmath241 has one nodal line , @xmath226 . folding this square gives indeed @xmath245 nodal domains if @xmath153 is odd , and @xmath248 nodal domains if @xmath153 is even .
this is illustrated ( as the right pictures ) in figures [ fig:3 - 0-special ] and [ fig:4 - 0-special ] .
, right @xmath249 . in each case
@xmath25 . note how the eigenfunction in the case @xmath250 is constructed by folding the ( scaled ) @xmath231 eigenfunction evenly.,title="fig:",width=113 ] , right @xmath249 . in each case
@xmath25 . note how the eigenfunction in the case @xmath250 is constructed by folding the ( scaled ) @xmath231 eigenfunction evenly.,title="fig:",width=113 ] with @xmath222 , right @xmath249 with @xmath225 .
both these cases give the maximal cardinal of @xmath251 nodal domains.,title="fig:",width=113 ] with @xmath222 , right @xmath249 with @xmath225 .
both these cases give the maximal cardinal of @xmath251 nodal domains.,title="fig:",width=113 ] with @xmath222 , right @xmath252 with @xmath225 .
note that @xmath222 gives @xmath253 nodal domains , which is the maximal cardinal , while @xmath225 gives @xmath254 nodal domains , only.,title="fig:",width=113 ] with @xmath222 , right @xmath252 with @xmath225 .
note that @xmath222 gives @xmath253 nodal domains , which is the maximal cardinal , while @xmath225 gives @xmath254 nodal domains , only.,title="fig:",width=113 ] [ cor : p0 ] the eigenvalues @xmath255 , @xmath256 , @xmath257 , @xmath258 , @xmath259 , @xmath102 , @xmath260 , @xmath261 , and @xmath144 are not courant sharp .
[ lem : pp ] if the eigenspace corresponding to @xmath262 is one - dimensional and @xmath34 is an eigenfunction corresponding this eigenvalue , then @xmath263 .
the eigenspace is spanned by @xmath264 , which is a true product of a function depending on @xmath265 and one that depends on @xmath266 .
each of them has @xmath153 zeros , and thus the number of nodal domains equals @xmath267 .
[ cor : pp ] the eigenvalues @xmath268 , @xmath269 , @xmath270 , @xmath136 , @xmath139 , @xmath178 , and @xmath149 are not courant sharp .
the reasoning below depends on the fact that the number of nodal domains of a continuous curve of eigenfunctions @xmath271 is constant unless there are interior stationary points appearing in the zero - set , i.e. @xmath272 such that @xmath273 or changes in the cardinal of the boundary points .
we refer for this point to lemma 4.4 in @xcite .
hence the analysis of these situations is quite important .
[ lemma5.1 ] if @xmath82 is odd , and @xmath274 , then @xmath275 satisfies at the point @xmath276 in @xmath3 if and only if @xmath277 if these two equations are fulfilled , we recover the critical value of @xmath201 via @xmath278 the eigenfunctions have the form @xmath279 the zero critical points are determined by @xmath280 since @xmath281 , the critical points @xmath276 should satisfy .
( note that one can reduce by dilation to the case when @xmath153 and @xmath154 are mutually prime . )
once a pair satisfying these two conditions is given , we recover the corresponding critical values of @xmath201 by .
the intersection of the zero set of @xmath282 with the boundary is determined by the equations @xmath283 outside the zeros of @xmath284 , the function @xmath285 has derivative @xmath286 is it easy from the graph of @xmath287 to count for a given @xmath201 the number of points arriving at the boundary .
except at the corners ( see remark [ remcor ] below ) this number is changing at the critical values of @xmath287 ( see remark [ rem4.4 ] ) , which are the solutions of @xmath288 in the case by case analysis we will always have in mind the following remarks . [ rem4.4 ] the critical points at the boundary ( except for the corners ) are given by @xmath289 , @xmath290 , @xmath291 and @xmath292 where @xmath293 is some non zero solution of .
the corresponding critical @xmath201 is obtained by equation .
[ remcor ] the values for which lines arrive to the corner correspond only to @xmath294 and @xmath295 .
[ rmcrtbnd ] the solutions of @xmath296 can also be obtained by looking at the local extrema of @xmath297 .
[ lmutuallyprime ] the analysis of the solutions of @xmath296 can ( by a change of variable ) be reduced to the case when @xmath153 and @xmath154 are mutually prime . from these remarks , we deduce that for a complete analysis of the nodal patterns corresponding to a pair @xmath166 such that the eigenvalue @xmath81 has multiplicity @xmath298 , we should first analyze the graph of the function @xmath297
. this will not only permit to count in function of @xmath201 the number of lines touching the boundary but will also permit to determine by the analysis of the local extrema to determine the critical value of @xmath201 for which we have critical points .
this idea was proposed by a. stern @xcite and used intensively and more rigorously in @xcite .
we consider a pair @xmath166 with @xmath166 mutually prime and @xmath82 odd and would like to localize the zeros of @xmath299 for say @xmath300 .
it is based on a very elementary observation .
we simply observe that if @xmath301 , then @xmath302 .
this determines the `` white '' rectangles of a chessboard .
these rectangles are obtained by drawing the vertical lines @xmath303 ( @xmath183 odd ) and @xmath304 ( @xmath305 odd ) , and similarly the horizontal lines @xmath306 ( @xmath183 odd ) and @xmath307 ( @xmath305 odd ) , and hence the zero set should be contained in the closed `` black '' rectangles corresponding to the closure of the set @xmath308 .
note that these rectangles have different size .
it is also important to determine which points at the boundary of a given rectangle belongs to the zero set .
they are obtained by the equations @xmath309 and for @xmath310 for @xmath183 , @xmath311 , @xmath305 , @xmath312 odd .
so the nodal set should contain all these points and only these points .
we call these points _ admissible corners_. this can also be seen as a consequence of the fact that @xmath313 and @xmath314 have no common zero in @xmath315 $ ] when @xmath82 is odd .
hence we have proved .
[ lemma5.6 ] if @xmath153 and @xmath154 are mutually prime and @xmath82 is odd , and @xmath300 then the only points of intersection of the zero set of @xmath299 with the boundary of a black rectangle are the admissible corners . moreover , these points are regular points of the zero set .
another point is that the zero set can not contain any closed curve inside a black rectangle ( we also mean curves touching the boundary ) .
the ground state energy inside the curve delimited by the curve ( say in the case without double points ) should indeed be strictly above the ground state energy of the rectangle ( dirichlet for a rectangle in the interior , dirichlet
neumann when the rectangle has at least one size common to the square @xmath315 ^ 2 $ ] ) .
but the minimal energy for these rectangles is @xmath316 ( a contradiction with the value @xmath317 ) .
[ lemma5.7 ] if @xmath153 and @xmath154 are mutually prime and @xmath82 is odd and @xmath300 then the zero set of @xmath299 can not contain any closed curve contained in a `` black '' rectangle . as a consequence of these two lemmas
let us observe that at an admissible corner only one curve belonging to the zero set can enter in a black rectangle and that it should either go out by an admissible corner , either touch the boundary or meet another curve of the zero set at a critical point .
most of the cases appearing in the table are treated via the general considerations of sections [ section2 ] and [ section3 ] . in this section ,
we consider a first list of special cases where a more careful analysis is needed , which involves the analysis of boundary points or of critical points .
the eigenspace is two - dimensional , @xmath320 we know from lemma [ lem : antisymmetric ] that this case is not courant sharp , but that it has a maximum number of nodal domains being @xmath321 .
[ lem:2 - 1 ] @xmath322 .
if @xmath281 then @xmath323 . observing that @xmath324 it is immediate that @xmath325 has no zero in the open interval @xmath326 . hence by lemma [ lemma5.1 ] , there are no critical points for @xmath327 in @xmath3 .
having in mind , the analysis of the graph @xmath328 ( see figure [ fig:2 - 1-cos ] ) leads immediately ( see figure [ fig:2 - 1 ] for @xmath329 ) to the existence of @xmath159 nodal domains in this case ( two boundary points for @xmath209 and @xmath210 and one boundary point for @xmath211 and @xmath212 ) . in the interval @xmath189.,width=302 ] when @xmath221 , @xmath330 , and @xmath222.,title="fig:",width=113 ] when @xmath221 , @xmath330 , and @xmath222.,title="fig:",width=113 ] when @xmath221 , @xmath330 , and @xmath222.,title="fig:",width=113 ] it remains to consider two cases , @xmath221 and @xmath331 .
for @xmath221 we are in the product case and have @xmath321 nodal domains .
for @xmath222 we let @xmath332 and @xmath333 , both living in @xmath334 $ ] . then @xmath335 thus , we get the straight line @xmath336 and the hyperbola @xmath337 .
we note that they do not intersect .
thus , there are @xmath159 nodal domains in this case .
[ lem:4 - 2 ] assume that @xmath340 is an eigenpair of @xmath9 .
then @xmath341 .
in particular we are not in the courant sharp situation . if @xmath342 , then lemmas [ lem : pandqeven ] and [ lem:2 - 1 ] give that the number of nodal domains is less than or equal to @xmath343 . for @xmath344
we are in the product case , and have @xmath345 nodal domains .
[ lem:8 - 4 ] assume that @xmath348 is an eigenpair of @xmath9
. then @xmath349 .
in particular we are not in the courant sharp situation . if @xmath342 , then lemmas [ lem : pandqeven ] and [ lem:4 - 2 ] imply that the number of nodal domains is less than or equal to @xmath350 . for @xmath344
we are in the product case , and have @xmath351 nodal domains .
[ lem:6 - 3 ] assume that @xmath354 is an eigenpair of @xmath9
. then @xmath355 .
in particular , the eigenpair @xmath354 is not courant sharp .
fix @xmath356 .
it holds that @xmath357 in the square @xmath358 .
moreover , the values of @xmath359 in the rest of @xmath3 are recovered by `` folding evenly '' .
thus , the number of nodal domains are less then or equal to ( in fact , this could be made sharper ) nine times the number of nodal domains of @xmath360 . from lemma [ lem:2 - 1 ]
it follows that if @xmath342 then @xmath361 if @xmath344 then we are in the product case and have @xmath362 nodal domains .
here the situation is similar to @xmath319 , but with one additional nodal domain touching each part of the boundary .
the general eigenfunction is @xmath365 [ lem:3 - 2 ] if @xmath221 then @xmath366 . if @xmath367 , @xmath368 . in particular , if @xmath369 is an eigenpair of @xmath9 , it is not courant sharp . by lemma [ lemma3.8 ]
it is sufficient to consider @xmath370 . if @xmath221 we are in the product case and there are exactly @xmath254 nodal domains .
assume that @xmath281 .
let us start to count the nodal lines that touch the boundary .
for this , we use with @xmath364 .
the function @xmath371 , @xmath372\setminus\{\pi/4,3\pi/4\}$ ] has derivative @xmath373 in the interval @xmath189.,width=302 ] in particular , @xmath374 is negative where it is defined .
we find immediately that @xmath375 attains all values in @xmath334 $ ] three times and all values in @xmath376\}$ ] twice .
using lemma [ lemma5.1 ] and remark [ rmcrtbnd ] and the fact that @xmath375 has no critical points in its domain of definition , we get that we have no interior critical points of @xmath377 .
we conclude ( see for example figure [ fig:3 - 2 ] for @xmath378 ) that if @xmath379 then we have exactly three nodal lines touching each part of the boundary of @xmath3 where @xmath209 and @xmath210 , and exactly two nodal lines on each of the parts of the boundary of @xmath3 where @xmath211 and @xmath212 .
thus , we only have to consider the case @xmath222 .
if @xmath222 then we have one nodal line @xmath223 and two additional nodal lines touching each part of the boundary .
next , we eliminate the case of loops in the zero set by using lemma [ lemma5.7 ] . in this simple case , we can do a more algebraic computation .
we make the substitution @xmath332 and @xmath333 and find that @xmath380 if and only if @xmath381 one solution is @xmath336 .
let @xmath382 .
then @xmath383 and hence @xmath384 it is easily seen that @xmath385 . next , @xmath386 and since @xmath387 we find that @xmath388 and @xmath389 have no common zeros .
thus , the nodal lines are never vertical . by symmetry in @xmath390 and @xmath391
it follows that they are never horizontal either .
all in all , this means that for @xmath222 ( and thus for @xmath274 ) we have five non - intersecting nodal lines , and so six nodal domains . when @xmath221 , @xmath330 , and @xmath392.,title="fig:",width=113 ] 0.5 cm when @xmath221 , @xmath330 , and @xmath392.,title="fig:",width=113 ] 0.5 cm when @xmath221 , @xmath330 , and @xmath392.,title="fig:",width=113 ] [ lem:6 - 4 ] if @xmath344 then @xmath395 . for all other values of @xmath396 , @xmath397 .
in particular , if @xmath398 is an eigenpair of @xmath399 , then it is not courant sharp . if @xmath344 then we are in the product case , and have @xmath400 nodal domains .
if @xmath342 then , combining lemmas [ lem : pandqeven ] and [ lem:3 - 2 ] , we find that the number of nodal domains are at most @xmath401 .
it remains to analyze the four cases @xmath402 , @xmath403 , @xmath404 , and @xmath405 .
we will start by a detailed analysis of the case @xmath402 . for the last three cases we will use a chessboard localization to improve the estimate @xmath406 used in section [ section2 ] ( see lemma [ lemma2.2 ] ) . [ lem:4 - 1 ] for any @xmath201 , @xmath409 .
the possible values of @xmath410 are @xmath321 , @xmath251 and @xmath411 .
in particular , if @xmath412 is an eigenpair of @xmath9 , then it is not courant sharp .
the case @xmath221 being very simple to analyze ( product situation , @xmath411 nodal domains , hence not courant sharp ) . by lemma [ lemma3.8 ] , it is sufficient to do the analysis for @xmath413 $ ] . [
[ step-1-analysis - of - the - graph - of - f_41 ] ] * step 1 : analysis of the graph of @xmath414 * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as explained in section [ section5 ] , everything can be read on the graph of @xmath414 . in the interval @xmath189.,width=302 ] from the graph , we find that @xmath414 attains all values in @xmath415 - 1,1[$ ] four times , and all values in @xmath376\}$ ] three , two or one times .
this transition has to be analyzed further .
looking at the local extrema of @xmath414 , there are two points @xmath416 and @xmath417 which are solutions for .
@xmath418 we can now follow the scheme of analysis presented in section [ section5 ] . [ [ step-2-interior - critical - points . ] ] * step 2 : interior critical points . * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + using the symmetry of @xmath416 and @xmath417 with respect to @xmath419 , we get that , for @xmath413 $ ] , the only stationary points of @xmath420 are @xmath421 and @xmath422 and appear when @xmath222 .
so we need a special analysis for @xmath423 . from figure
[ fig:4 - 1 ] we see that the number of nodal domains is @xmath411 .
we immediately see that the anti - diagonal belongs to the zero set .
[ [ step-3-analysis - of - the - boundary - points ] ] * step 3 : analysis of the boundary points * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + here we have to use for @xmath408 and our preliminary analysis of the graph of @xmath414 .
we conclude that if @xmath379 then we have exactly four nodal lines touching each part of the boundary of @xmath3 where @xmath209 and @xmath210 , but the number of nodal lines on each of the parts of the boundary of @xmath3 where @xmath211 and @xmath212 can be @xmath424 , @xmath298 or @xmath425 .
there is another critical value @xmath426 corresponding to a local maximum of @xmath414 : @xmath427 one touching occurs at @xmath428 and simultaneously at @xmath429 .
it remains to count the number of nodal domains in this situation and to count the number of nodal domains for one value of @xmath430 .
in each of these intervals two strategies are possible : * analyze perturbatively the situation for @xmath201 close to one of the ends of the interval ; * choose one specific value of @xmath201 in the interval . in our case , we have finally three critical values of @xmath201 ( @xmath431 , @xmath432 and @xmath433 ) and two values to choose in @xmath434 and @xmath435 . because @xmath436 , the picture for @xmath330 in figure [ fig:4 - 1 ] is the answer in the interval @xmath434 .
hence , we see six nodal domains when @xmath437 . it remains to analyze the situation for @xmath438 and to analyze another case in @xmath435 .
we observe @xmath251 nodal domains for @xmath439 and come back to @xmath411 for @xmath294 . but
this is the @xmath440-th eigenvalue .
[ [ step-4-chessboard - localization ] ] * step 4 : chessboard localization * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we refer to section [ section5 ] and particularly to the lemmas [ lemma5.6 ] and [ lemma5.7 ] .
we now consider @xmath201 in the interval @xmath434 .
we know that there are no critical points inside .
hence one line entering in a rectangle by one of the admissible corner belonging to the nodal set should exit the black rectangle by another corner or by the boundary .
conversely , a line starting from the boundary should leave the black rectangle through a corner in the zero set .
note that contrary to the case considered in @xcite it is not true that all the corners belong to the zero set .
we now look at the points on the boundary . for @xmath211
, we have shown that there is only one point @xmath441 .
moreover @xmath442 .
similar considerations can be done to localize the four points on @xmath209 and on @xmath210 .
these localization are independent of @xmath443 .
finally , we notice that @xmath444 meets the nodal set at a unique point @xmath445 and the same for @xmath446 .
it is then easy to verify that one can reconstruct uniquely the nodal picture using these rules . for @xmath438 and @xmath430 , similar arguments lead to a unique topological type . for @xmath294
, one should first analyze the neighborhood of the anti - diagonal where critical points are present .
, @xmath447.,width=226 ] , @xmath448.,width=226 ] [ lem:8 - 3 ] let @xmath451 be an eigenpair of @xmath9
. then @xmath452 . in particular
, @xmath451 is not courant sharp . by lemma [ lemma3.8 ]
it is sufficient to estimate the nodal domains of @xmath453 for @xmath454 .
first , we note that for @xmath221 we are in the product situation and have @xmath455 nodal domains .
next , we use the chessboard argument . for all @xmath281
it holds that the function @xmath456 in the white rectangles . thus no nodal lines can cross white rectangles
moreover , since both @xmath457 and @xmath458 , we find that the nodal lines must pass corners where both @xmath459 and @xmath460 are zero , and that they can not pass corners where only one of them are zero ( marked with a red cross ) . in figure
[ fig:8 - 3-chess - pi4-marked ] , we paint white rectangles blue in the following way : first we let each white rectangle touching the boundary become blue .
then we paint each white rectangle having a forbidden corner ( marked with a red cross ) in common with a blue rectangle .
this latter procedure is repeated until it does not apply anymore .
the so recolored blue rectangles are then necessarily all subsets of @xmath40 of nodal domains touching the boundary .
note that this construction is independent of @xmath201 .
thus , @xmath461 = \tfrac{3}{8}\pi^2,\ ] ] and hence @xmath462 then lemma [ lemma2.2 ] gives , @xmath463 , @xmath464.,width=302 ] [ lem:9 - 4 ] let @xmath467 be an eigenpair of @xmath9
. then @xmath468 .
in particular , @xmath467 is not courant sharp .
this proof goes line by line as the proof of lemma [ lem:8 - 3 ] , with a change of numbers only .
for @xmath221 we are in the product situation , with @xmath469 nodal domains . here , the blue area of figure [ fig:9 - 4-chess - pi4-marked ] equals @xmath470 , and so we know that @xmath40 has this lower bound .
this means that @xmath471 we get from lemma [ lemma2.2 ] @xmath472 , @xmath473.,width=302 ] [ lem:10 - 4 ] let @xmath476 be an eigenpair of @xmath9
. then @xmath477 . in particular
, @xmath476 is not courant sharp .
we first note that we are in the product situation if @xmath221 or @xmath224 , with @xmath478 nodal domains .
since @xmath479 and @xmath480 have no common zeros , we can again apply the chessboard argument for @xmath481 .
for @xmath482 the argument is exactly the same , but with the roles of white and black rectangles interchanged .
thus , as in the previous proofs , we count the area of the blue region , see figure [ fig:10 - 4-chess - pi4-marked ] . we find that it equals @xmath483 .
hence @xmath484 , and by lemma [ lemma2.2 ] @xmath485 , @xmath473.,width=302 ] with the proof of this last statement and the analysis presented in the table we have achieved the proof of theorem [ thm : main ] .
in the next section we will by curiosity analyze the spectral pattern of some of the families .
although not used in the proof of the main results in this paper , we present in the spirit of the analysis of the case @xmath402 a complete analysis of the nodal pattern in the cases @xmath486 and @xmath403 .
we already know that we are not courant sharp by lemma [ lem : antisymmetric ] .
it does not cost too much ( an application could be for the analysis of the case @xmath405 but we only need a weaker upper bound ) to establish the [ lem:5 - 2 ] for any @xmath201 , we have the optimal bound @xmath489 .
the analysis of the equation @xmath490 leads to the existence of two positive solutions @xmath421 of this equation with @xmath491 these values appear also as the critical values of @xmath492 as can be seen in figure [ fig:5 - 2-cos ] . in the interval @xmath189.,width=302 ] it is sufficient to analyze the situation for @xmath413 $ ] .
the discussion is rather close to the case @xmath402 . for @xmath493 , we start from @xmath494 nodal domains .
we have indeed @xmath411 critical points and @xmath495 boundary points ( avoiding the corners ) . in the interval @xmath496 , there are no critical point inside the square .
but there are transition at the boundary for @xmath497 such that @xmath498 .
hence the number of nodal domains is fixed in @xmath434 and because when starting from @xmath431 , we have only opening at the former crosses , the number of nodal domains can only decrease .
more precisely , the number of nodal domain is eight .
this results from the numerics or the perturbative analysis starting from @xmath221 .
an analysis for @xmath499 should be done .
then again the number of nodal domains is fixed in @xmath435 and equal to @xmath411 .
at @xmath423 , the nodal set contains the anti - diagonal @xmath500 .
the number of boundary points outside the corners become @xmath159 on each side .
we have two critical points @xmath421 and @xmath422 on the anti - diagonal .
numerics permits us to determine the number of nodal domains , which increases by @xmath298 and is equal to @xmath411 for @xmath294 .
we keep in mind the results established in section [ section5 ] on the chessboard localization .
we now consider @xmath201 in the interval @xmath434 . we know that there are no critical points inside .
hence one line entering in a rectangle by one of the corners belonging to the nodal set should exit the black rectangle by another corner or by the boundary .
conversely , a line starting from the boundary should leave the black rectangle through a corner in the zero set .
note that contrary to the case considered in @xcite it is not true that all the corners belong to the zero set .
we now look at the points on the boundary . for @xmath211
, we have shown that there are exactly two points @xmath501 and @xmath502 .
moreover @xmath503 and @xmath504 .
similar considerations can be done to localize the five points on @xmath209 , @xmath505 , and on @xmath210 , @xmath506 , and two points on @xmath212 , @xmath507 .
these localizations are independent of @xmath443 .
let us see if one can reconstruct uniquely the ( topology of the ) nodal picture using these rules .
+ the nodal line starting from the boundary at @xmath501 has no choice ( that is the ordered sequence of admissible corners which are visited is uniquely determined ) and should arrive to @xmath508 .
the line starting of @xmath502 has no choice and should arrive to @xmath509 .
the line starting from @xmath510 should come back to @xmath511 .
the line starting from @xmath512 is obliged to go to @xmath513 and the line starting from @xmath514 has to go to @xmath515 .
all these lines are unique .
it remains one line which has to go from @xmath516 to @xmath517 with the obligation to visit all the elements of the two lattices which have not been visited before .
a small analysis shows that it remains two possible paths around the center ( see figure [ fig:5 - 2alt ] ) .
hence we need an additional argument to fix the sequence of visited admissible corners .
for example it is enough to show that on the line @xmath518 there are no zero @xmath519 with @xmath520 .
this is at least clear for @xmath201 small and because no critical point can appear before @xmath423 .
we are done with this case . ,
@xmath521.,width=302 ] , @xmath448.,width=302 ] for @xmath430 , similar arguments lead to a unique topological type .
we have now four points @xmath522 ( @xmath523 ) and four points @xmath524 ( @xmath523 ) at the vertical boundaries but except a change in the black rectangle containing @xmath502 and @xmath525 and the black rectangle containing @xmath515 and @xmath526 , nothing changes outside . at a first sight , there are still two possibilities but the transition to the second possibility can only occur through a critical point .
this is impossible before @xmath294 . as a corollary
, we recover that the eigenvalue @xmath527 is not courant sharp .
this is an immediate consequence of lemma [ lem : pandqeven ] and lemma [ lem:5 - 2 ] .
this gives : @xmath528 this can be improved by using the detailed case by case analysis of @xmath529 .
we will then get the optimal upper - bound : @xmath530 to compare with lemma [ lem:10 - 4 ] .
[ lem:8 - 3-opt ] for any @xmath201 , we have the optimal bound @xmath531 . in the interval @xmath189.,width=302 ] the analysis of the figure [ fig:8 - 3-cos ] shows the existence of six critical points : @xmath532 with @xmath533 we associate with these critical values the two positive numbers : @xmath534 and observe that @xmath535 associated with @xmath536 we introduce the two critical angles in @xmath496 : @xmath537 for these two values some transition should appear at the boundary .
we now look at the interior critical points corresponding to pairs @xmath538 ( @xmath539 and @xmath540 .
the corresponding critical @xmath541 are determined by @xmath542 with @xmath543 . using the symmetries ,
it is enough to look at the ones which belong to @xmath544 $ ] .
we recover @xmath433 with any pair @xmath545 but we have also to consider @xmath546 which is determined by @xmath547 .
we observe ( numerically ) that @xmath548 hence we have at the end to look at the values @xmath431 , @xmath432 , @xmath549 , @xmath550 and @xmath433 and to four values of @xmath201 corresponding to each of the intervals @xmath551 , @xmath552 , @xmath553 and @xmath554 .
we keep in mind the results obtained in section [ section5 ] concerning the chessboard localization and its consequences .
+ we now consider @xmath201 in the interval @xmath434 .
we know that there are no critical points inside .
hence one line entering in a rectangle by one of the corner belonging to the nodal set should exit the black rectangle by another corner or by the boundary .
conversely , a line starting from the boundary should leave the black rectangle through a corner in the zero set .
we now look at the points on the boundary . for @xmath211 ,
we have shown that there are exactly three points @xmath501 , @xmath502 , and @xmath525 .
moreover @xmath555 , @xmath556 , and @xmath557 .
similar considerations can be done to localize the nine points on @xmath209 , @xmath558 , and on @xmath210 , @xmath559 , and three points on @xmath212 , @xmath560 .
these localizations are independent of @xmath443 .
let us see if one can reconstruct uniquely the nodal picture using these rules .
the nodal line starting from @xmath501 has no other choice than going through one admissible corner to the point @xmath508 .
the nodal line starting from @xmath502 has no other choice than going to @xmath509 after passing through five admissible corners .
the curve starting from @xmath561 has no other choice than coming back to the same boundary at @xmath562 after passing through two admissible corners .
similarly , the line starting from @xmath563 has no other choice than coming back to the same boundary at @xmath564 after passing through two admissible corners . for the nodal line starting from @xmath525 , the first five admissible corners to visit are uniquely determined by the given rules .
then the line enters in a rectangle with four admissible corners .
there are two choices for leaving this rectangle .
the determination of the right admissible corner can be done by using perturbation theory or a barrier argument .
this leads to go down to the left down corner . after visiting this one the two next admissible corners are uniquely determined .
the nodal line enters in a rectangle with four admissible corners .
again , we have to use a perturbation argument to decide that we have to leave at the admissible left up corner . then everything is uniquely determined till the nodal line touches the boundary at @xmath517 .
we now use the symmetry with respect to the diagonal to draw three new nodal lines .
the last line joining @xmath516 to @xmath565 is then uniquely determined . in this way
we get twelve nodal domains .
the case @xmath499 corresponds to a change at the boundary . instead of three lines touching at the boundary at @xmath211 and @xmath212 , two new lines touch the boundary at the same point at @xmath211 between the former @xmath501 and @xmath502 ( resp at @xmath212 between @xmath515 and @xmath526 ) .
the number of nodal domains becomes equal to @xmath495 . for @xmath566 ,
nothing has changed except that we have now exactly five points at @xmath211 and five points at @xmath212 .
the number of nodal domains is constant and equal to @xmath495 . for @xmath567 , two critical points appear inside the square leading to the creation of two new nodal domains .
we have now sixteen nodal domains .
nothing has changed at the boundary . for @xmath568 ,
the two critical points disappear .
nothing changes at the boundary and we keep @xmath569 nodal domains . for @xmath570 , two new lines touch the boundary at the same point at @xmath211 and similarly at @xmath212 .
this creates two new nodal domains .
we now get @xmath440 nodal domains . for @xmath571 , we have now seven touching points at @xmath211 and seven at @xmath212 .
the number of nodal domains remains equal to @xmath440 . finally , for @xmath294 , four critical points appear on the anti - diagonal .
four nodal domains are created .
we have now @xmath572 nodal domains .
this ends the ( sketch of ) the proof of proposition [ lem:8 - 3-opt ]
. 0.5 cm 0.5 cm
cccccccc 1 & 0 & 0 & 0 & & 1 & & courant sharp + 23 & 1 & 0 & 1 & 12 & & & courant sharp + & 0 & 1 & & & & & + 4 & 1 & 1 & 2 & & 2 & 1 & courant sharp + 56 & 2 & 0 & 4 & & 34 & & courant sharp + & 0 & 2 & & & & & + 78 & 2 & 1 & 5 & 34 & & & lemma [ lem : antisymmetric ] + & 1 & 2 & & & & & + 9 & 2 & 2 & 8 & & 5 & & courant sharp + 1011 & 3 & 0 & 9 & 56 & & & lemma [ lem : p0 ] + & 0 & 3 & & & & & + 1213 & 3 & 1 & 10 & & 67 & 23 & lemma [ lem : antimirror ] + & 1 & 3 & & & & & + 1415 & 3 & 2 & 13 & 78 & & & lemma [ lem:3 - 2 ] + & 2 & 3 & & & & & + 1617 & 4 & 0 & 16 & & 89 & & lemma [ lem : p0 ] + & 0 & 4 & & & & &
+ 1819 & 4 & 1 & 17 & 910 & & & lemma [ lem:4 - 1 ] + & 1 & 4 & & & & & + 20 & 3 & 3 & 18 & & 10 & 4 & lemma [ lem : pp ] + 2122 & 4 & 2 & 20 & & 1112 & & lemma [ lem:4 - 2 ] + & 2 & 4 & & & & & + 2326 & 5 & 0 & 25 & 1114 & & & lemma [ lem : antisymmetric ] + & 4 & 3 & & & & & + & 3 & 4 & & & & & + & 0 & 5 & & & & & + 2728 & 5 & 1 & 26 & & 1314 & 56 & lemma [ lem : antimirror ] + & 1 & 5 & & & & & + 2930 & 5 & 2 & 29 & 1516 & & & lemma [ lem : antisymmetric ] + & 2 & 5 & & & & & + 31 & 4 & 4 & 32 & & 15 & & lemma [ lem : pp ] + 3233 & 5 & 3 & 34 & & 1617 & 78 & lemma [ lem : antimirror ] + & 3 & 5 & & & & & + 3435 & 6 & 0 & 36 & & 1819 & & lemma [ lem : p0 ] + & 0 & 6 & & & & & + 3637 & 6 & 1 & 37 & 1718 & & & lemma [ lem : antisymmetric ] + & 1 & 6 & & & & & + 3839 & 6 & 2 & 40 & & 2021 & & lemma [ lem : pandqeven ] + & 2 & 6 & & & & & + 4041 & 5 & 4 & 41 & 1920 & & & lemma [ lem : antisymmetric ] + & 4 & 5 & & & & & + 4243 & 6 & 3 & 45 & 2122 & & & lemma [ lem:6 - 3 ] + & 3 & 6 & & & & & + 4445 & 7 & 0 & 49 & 2324 & & & lemma [ lem : p0 ] + & 0 & 7 & & & & & + 4648 & 7 & 1 & 50 & & 2224 & 911&lemma [ lem : antimirror ] + & 5 & 5 & & & & & + & 1 & 7 & & & & & + 4950 & 6 & 4 & 52 & & 2526 & & lemma [ lem:6 - 4 ] + & 4 & 6 & & & & & + 5152 & 7 & 2 & 53 & 2526 & & & lemma [ lem : antisymmetric ] + & 2 & 7 & & & & & + 5354 & 7 & 3 & 58 & & 2728&1213 & lemma [ lem : antimirror ] + & 3 & 7 & & & & + 5556 & 6 & 5 & 61 & 2728 & & & lemma [ lem : antisymmetric ] + & 5 & 6 & & & & & + 5758 & 8 & 0 & 64 & & 2930 & & lemma [ lem : p0 ] + & 0 & 8 & & & & & + 5962 & 8 & 1 & 65 & 2932 & & & lemma [ lem : antisymmetric ] + & 7 & 4 & & & & & + & 4 & 7 & & & & & + & 1 & 8 & & & & & + 6364 & 8 & 2 & 68 & & 3132&&lemma [ lem : symmetric ] + & 2 & 8 & & & & & + 65 & 6 & 6 & 72 & & 33 & & lemma [ lem : pp ] + 6667 & 8 & 3 & 73 & 3334 & & & lemma [ lem:8 - 3 ] + & 3 & 8 & & & & & + 6869 & 7 & 5 & 74 & & 3435 & 1415 & lemma [ lem : antimirror ] + & 5 & 7 & & & & & + 7071 & 8 & 4 & 80 & & 3637 & & lemma [ lem:8 - 4 ] + & 4 & 8 & & & & & + 7273 & 9 & 0 & 81 & 3536 & & & lemma [ lem : p0 ] + & 0 & 9 & & & & & + 7475 & 9 & 1 & 82 & & 3839 & 1617 & lemma [ lem : antimirror ] + & 1 & 9 & & & & & + 7679 & 9 & 2 & 85 & 3740 & & & lemma [ lem : antisymmetric ] + & 7 & 6 & & & & & + & 6 & 7 & & & & & + & 2 & 9 & & & & & + 8081 & 8 & 5 & 89 & 4142 & & 1819 & lemma [ lemma2.3 ] + & 5 & 8 & & & & & + 8283 & 9 & 3 & 90 & & 4041 & 2021 & lemma [ lem : antimirror ] + & 3 & 9 & & & & & + 8485 & 9 & 4 & 97 & 4344 & & & lemma [ lem:9 - 4 ] + & 4 & 9 & & & & & + 86 & 7 & 7 & 98 & & 42 & 22 & lemma [ lem : antimirror ] + 8790 & 10 & 0 & 100 & & 4346 & & lemma [ lem : symmetric ] + & 8 & 6 & & & & & + & 6 & 8 & & & & & + & 0 & 10 & & & & & + 9192 & 10 & 1 & 101 & 4546 & & & lemma [ lem : antisymmetric ] + & 1 & 10 & & & & & + 9394 & 10 & 2 & 104 & & 4748 & & lemma [ lem : pandqeven ] + & 2 & 10 & & & & & + 9596 & 9 & 5 & 106 & & 4950 & 2324 & lemma [ lemma2.3 ] + & 5 & 9 & & & & & + 9798 & 10 & 3 & 109 & 4748 & & & lemma [ lem : antisymmetric ] + & 3 & 10 & & & & & + 99100 & 8 & 7 & 113 & 4950 & & & lemma [ lemma2.3 ] + & 7 & 8 & & & & & + 101102 & 10 & 4 & 116 & & 5152 & & lemma [ lem:10 - 4 ] + & 4 & 10 & & & & & + 103104 & 9 & 6 & 117 & 5152 & & & lemma [ lemma2.3 ] + & 6 & 9 & & & & & + 105106 & 11 & 0 & 121 & 5354 & & & lemma [ lem : p0 ] + & 0 & 11 & & & & & + 107108 & 11 & 1 & 122 & & 5354 & 2526 & lemma [ lem : antimirror ] + & 1 & 11 & & & & & + 109112 & 11 & 2 & 125 & 5558 & & & lemma [
lem : antisymmetric ] + & 10 & 5 & & & & & + & 5 & 10 & & & & & + & 2 & 11 & & & & & + 113 & 8 & 8 & 128 & & 55 & & lemma [ lemma2.3 ] + 114117 & 11 & 3 & 130 & & 5659 & 2730 & lemma [ lem : antimirror ] + & 9 & 7 & & & & & + & 7 & 9 & & & & & + & 3 & 11 & & & & & + 118119 & 10 & 6 & 136 & & 6061 & & lemma [ lemma2.3 ] + & 6 & 10 & & & & & + 120121 & 11 & 4 & 137 & 5960 & & & lemma [ lemma2.3 ] + & 4 & 11 & & & & & + 122123 & 12 & 0 & 144 & & 6263 & & lemma [ lem : p0 ] + & 0 & 12 & & & & & + 124127 & 12 & 1 & 145 & 6164 & & & lemma
[ lem : antisymmetric ] + & 9 & 8 & & & & & + & 8 & 9 & & & & & + & 1 & 12 & & & & & + 128129 & 11 & 5 & 146 & & 6465 & 3132 & lemma [ lemma2.3 ] + & 5 & 11 & & & & & + 130131 & 12 & 2 & 148 & & 6667 & & lemma [ lemma2.3 ] + & 2 & 12 & & & & & + 132133 & 10 & 7 & 149 & 6566 & & & lemma [ lemma2.3 ] + & 7 & 10 & & & & & + 134135 & 12 & 3 & 153 & 6768 & & & lemma [ lemma2.3 ] + & 3 & 12 & & & & & + 136137 & 11 & 6 & 157 & 6970 & & & lemma [ lemma2.3 ] + & 6 & 11 & & & & & + 138139 & 12 & 4 & 160 & & 6869 & & lemma [ lemma2.3 ] + & 4 & 12 & & & & & + 140 & 9 & 9 & 162 & & 70 & 33 & lemma [ lemma2.3 ] + 141142 & 10 & 8 & 164 & & 7172 & & lemma [ lemma2.3 ] + & 8 & 10 & & & & & + 143146 & 13 & 0 & 169 & 7174 & & & lemma [ lem : antisymmetric ] + & 12 & 5 & & & & & + & 5 & 12 & & & & & + & 0 & 13 & & & & & + 147150 & 13 & 1 & 170 & & 7376 & 3437 & lemma [ lem : antimirror ] + & 11 & 7 & & & & & + & 7 & 11 & & & & & + & 1 & 13 & & & & & + 151152 & 13 & 2 & 173 & 7576 & & & lemma [ lemma2.3 ] + & 2 & 13 & & & & & + 153154 & 13 & 3 & 178 & & 7778 & 3839 & lemma [ lemma2.3 ] + & 3 & 13 & & & & & + 155156 & 12 & 6 & 180 & & 7980 & & lemma [ lemma2.3 ] + & 6 & 12 & & & & & + 157158 & 10 & 9 & 181 & 7778 & & & lemma [ lemma2.3 ] + & 9 & 10 & & & & & + 159162 & 13 & 4 & 185 & 7982 & & & lemma [ lem : antisymmetric ] + & 11 & 8 & & & & & + & 8 & 11 & & & & & + & 4 & 13 & & & & & + 163164 & 12 & 7 & 193 & 8384 & & & lemma [ lemma2.3 ] + & 7 & 12 & & & & & + 165166 & 13 & 5 & 194 & & 8182 & 4041 & lemma [ lemma2.3 ] + & 5 & 13 & & & & & + 167168 & 14 & 0 & 196 & & 8384 & & lemma [ lem : p0 ] + & 0 & 14 & & & & & + 169170 & 14 & 1 & 197 & 8586 & & & lemma [ lemma2.3 ] + & 1 & 14 & & & & & + 171173 & 14 & 2 & 200 & & 8587 & & lemma [ lemma2.3 ] + & 10 & 10 & & & & & + & 2 & 14 & & & & & + 174175 & 11 & 9 & 202 & & 8889 & 4243 & lemma [ lemma2.3 ] + & 9 & 11 & & & & & + 176179 & 14 & 3 & 205 & 8790 & & & lemma [ lemma2.3 ] + & 13 & 6 & & & & & + & 6 & 13 & & & & & + & 3 & 14 & & & & & + 180181 & 12 & 8 & 208 & & 9091 & & lemma [ lemma2.3 ] + & 8 & 12 & & & & & + 182183 & 14 & 4 & 212 & & 9293 & & lemma [ lemma2.3 ] + & 4 & 14 & & & & & + 184185 & 13 & 7 & 218 & & 9495 & 4445 & lemma [ lemma2.3 ] + & 7 & 13 & & & & &
+ 186189 & 14 & 5 & 221 & 9194 & & & lemma [ lemma2.3 ] + & 11 & 10 & & & & & + & 10 & 11 & & & & & + & 5 & 14 & & & & & + 190193 & 15 & 0 & 225 & 9699 & & & lemma [ lemma2.3 ] + & 12 & 9 & & & & & + & 9 & 12 & & & & & + & 0 & 15 & & & & & + 194195 & 15 & 1 & 226 & & 9596 & 4647 & lemma [ lemma2.3 ] + & 1 & 15 & & & & & + 196197 & 15 & 2 & 229 & 99100 & & & lemma [ lemma2.3 ] + & 2 & 15 & & & & & + 198199 & 14 & 6 & 232 & & 9798 & & lemma [ lemma2.3 ] + & 6 & 14 & & & & & + 200201 & 13 & 8 & 233 & 101102 & & & lemma [ lemma2.3 ] + & 8 & 13 & & & & & + 202203 & 15 & 3 & 234 & & 100101 & 4849 & lemma [ lemma2.3 ] + & 3 & 15 & & & & & + 204205 & 15 & 4 & 241 & 103104 & & & lemma [ lemma2.3 ] + & 4 & 15 & & & & & + 206 & 11 & 11 & 242 & & 102 & 50 & lemma [ lemma2.3 ] + 207208 & 12 & 10 & 244 & & 103104&&lemma [ lemma2.3 ] + & 10 & 12 & & & & &
+ @xmath573 & & & @xmath574 & & & & proposition [ prop : red1 ] +
from the numerics together with our mathematical analysis for specific eigenspaces , it seems reasonable to correct some traveling folk conjecture into the following one : in a given eigenspace of dimension @xmath298 , the maximal number of nodal domains is obtained for at least one eigenfunction @xmath299 for some @xmath575 .
the numerical work of corentin lna @xcite devoted to the analysis of spectral minimal partitions ( showing non nodal @xmath183-minimal partitions starting from @xmath576 ) suggests that there are only two courant sharp situations .
the case of the isotropic torus has finally been solved quite recently by c. lena @xcite . following the strategy of .
pleijel @xcite , his proof is a combination of a lower bound ( la weyl ) of the counting function with an explicit remainder term and of a faber
krahn inequality for connected domains on the torus with an explicit upper bound on the area .
it is also natural to ask if there are similar results to the results concerning the dirichlet problem on the square considered by a. stern and brard helffer , that is the existence of an infinite sequence of eigenvalues such that a corresponding eigenfunction has only two nodal domains .
we conjecture that it is impossible to find such a sequence . to justify this guess , one can try to show that in the neumann case , the number of nodal lines touching the boundary tends to @xmath577 as the eigenvalue tends to @xmath577 .
this has a nice connection with a recent result of t. hoffmann - ostenhof @xcite , saying that the only eigenfunction whose nodal set does not touch the boundary is the constant one . at the moment
, we can only prove the following : [ prop : touchingpoints ] let @xmath578 with @xmath579 .
then , for any @xmath580 $ ] the nodal lines of the eigenfunction @xmath282 have at least @xmath581 touching points at the boundary .
we first prove the proposition , with the additional assumption that @xmath582 and @xmath583 have no common zeros in @xmath584 $ ] .
as we have seen in subsection [ ss5.2 ] , the analysis of the zeros on the boundary is immediately related with the investigation of the solutions of @xmath585 with @xmath586 or @xmath587 .
the result is then a consequence of the following lemma : let @xmath578 with @xmath579 .
suppose that @xmath582 and @xmath583 have no common zeros in @xmath584 $ ] .
then , for any @xmath588 $ ] , the equation @xmath589 has exactly @xmath153 solutions in @xmath315 $ ] .
+ * we first observe that there are at most @xmath153 solutions .
* indeed , if we choose @xmath332 as new variable , we obtain a polynomial equation in the variable @xmath390 of degree @xmath153 , @xmath590 where @xmath591 is some chebyshev polynomial defined by @xmath592 hence we get our first observation ( counting with multiplicity ) . * we now show that there are at least @xmath153 solutions . * for @xmath593 ,
the solutions are the zeros of @xmath594 , that is @xmath595 the zeros @xmath596 ( @xmath597 ) of @xmath284 will play an important role .
we will indeed look at the function @xmath287 introduced in and they correspond to vertical asymptotes of the graph of @xmath287 . for @xmath598 ,
say @xmath599 , we have now to count the number of solutions of @xmath600 .
first we observe that there is ( at least ) one solution in @xmath601 and no solution in @xmath602 .
moreover @xmath287 is finite there .
we now consider the equation in the interval @xmath603 , for some @xmath604 . for a given interval
there are three cases . 1 .
there is a zero @xmath596 in @xmath605 . in this case
the range of @xmath287 always contains @xmath606 in particular there is always at least one point such that @xmath607 .
2 . there are no zeros of @xmath583 in @xmath605 and @xmath608 .
we observe in this case that at @xmath609 which belongs to @xmath605 , we have @xmath610 we will see below that the inequality is strict when @xmath153 and @xmath154 are mutually prime . in the case when we have equality , we have @xmath611 and we get from that we are at a local extremum of @xmath287 .
3 . there are no zeros of @xmath583 in @xmath605 and @xmath612 . in this case
the guess is that there are no zeros .
we will get it at the end of the argument but the information is not needed for our lower bound . to complete the lower bound we have simply to verify that for two intervals @xmath605 , @xmath613 not containing a zero of @xmath583 we are either in a sequence case ( 2 ) , case ( 3 ) or in a sequence case ( 3 ) , case ( 2 ) .
this implies that we have for @xmath614 at least two solutions in the union of the two intervals .
for @xmath615 the argument is the same if the inequality is strict in and we have a double point if there is an equality ( this will correspond to a critical point at the boundary ) .
if one of the intervals , say @xmath613 , contains a zero of @xmath583 , we play the same game as before but with the pair @xmath605 , @xmath616 . summing up
we get the lower bound by @xmath153 .
hence we have exactly @xmath153 zeros .
we can now finish the proof of the proposition under the additional assumption that @xmath582 and @xmath583 have no common zeros in @xmath584 $ ] .
the lemma can be applied ( depending on @xmath201 ) either to @xmath211 and @xmath212 or to @xmath209 and @xmath210 .
for the two other cases , we can use that for @xmath617 , @xmath600 has at least one solution . before attacking the general case , note the following lemma : if @xmath153 and @xmath154 are mutually prime , and @xmath618 , @xmath619 let us assume @xmath183 odd .
we want to show that @xmath620 . by contradiction
, we would have @xmath621 this can be written in the form @xmath622 by assumption , @xmath153 and @xmath154 are mutually prime .
this leads to @xmath623 , @xmath624 for an integer @xmath625 .
we have now to remember that @xmath626 .
this gives a contradiction .
let us now assume @xmath183 even .
we want to show that @xmath627 . by contradiction
, we would have @xmath628 @xmath153 and @xmath154 being mutually prime .
this leads to @xmath629 .
this is again impossible because @xmath630 .
this implies equality in all the lower bounds of the second part .
this implies also @xmath631 local extrema in @xmath326 for @xmath287 as can be seen in figures [ fig:5 - 2-cos ] and [ fig:8 - 3-cos ] .
we now explain how we can relax the assumption that @xmath582 and @xmath583 ( or equivalently @xmath632 and @xmath633 ) have no common zeros in @xmath315 $ ] ) .
a simple example is @xmath634 and @xmath635 , where @xmath636 is a common zero of @xmath637 and @xmath638 .
this zero is common to the all family @xmath639 .
looking at @xmath640 , we observe that we can regularize it at @xmath419 and that it is enough to apply the previous argument for showing that there is at least two solutions of @xmath641 for @xmath642 ( see figure [ fig : cos-3 - 1 ] ) .
this is of course trivial in this case . in the interval @xmath189.,width=302 ] the general case is similar .
we first determine the cardinal @xmath643 of the set @xmath644 in @xmath645 ) of the @xmath183 s such that @xmath646 is a zero of @xmath647 .
similarly , we consider the set of the @xmath311 such that @xmath648 for some @xmath183 .
another way of presentation is to claim the existence of a polynomial @xmath649 of degree @xmath643 such that @xmath650 so that the regularization of @xmath287 is given by : @xmath651 it suffices to do the same proof as before but keeping only in our construction the @xmath652 which are not the @xmath643 common zeros of @xmath582 and @xmath583 .
the proof is then identical and will give at least @xmath653 solutions of @xmath654 , hence of @xmath655 in addition to the @xmath643 previously obtained .
let @xmath578 with @xmath579 and @xmath656 $ ] .
then the eigenfunction @xmath657 satisfies @xmath658 this is obtained by using euler s formula , which implies ( forgetting the contribution of the critical points inside the square ) @xmath659 where @xmath660 is the number of connected components of @xmath661 ( @xmath662 being the zero set of @xmath34 ) . using @xmath663 and proposition [ prop : touchingpoints ] we get .
this gives a rather explicit way to prove that for a specific family which seems relatively generic ( any family corresponding to eigenvalues of multiplicity at most @xmath298 ) the number of nodal domains tends to @xmath577 .
hence we conjecture : for any sequence of eigenfunctions of the neumann problem in the square associated with an infinite sequence of eigenvalues , the number of nodal domains tends to @xmath577 .
the authors would like to thank p. brard , c. lna , j. leydold , t. hoffmann - ostenhof for remarks or transmission of information .
f. pockels .
ber die partielle differentialgleichung @xmath664 and deren auftreten in mathematischen physik . historical math .
cornell university .
( 1891 ) available at https://archive.org/details/berdiepartiell02pockuoft .
bemerkungen ber asymptotisches verhalten von eigenwerten und eigenfunctionen .
gttingen 1925 .
extracts and annotations available at http://www-fourier.ujf-grenoble.fr/~pberard/r/stern-1925-thesis-partial-reprod.pdf .
|
. pleijel has proved that in the case of the laplacian on the square with neumann condition , the equality in the courant nodal theorem ( courant sharp situation ) can only be true for a finite number of eigenvalues .
we identify five courant sharp eigenvalues for the neumann laplacian in the square , and prove that there are no other cases .
|
the properties of excited nuclei in thermal equilibrium are of great physical interest .
we are interested in excitation energies below some ( 400500 ) mev , where we may still describe the nucleus as a system of neutrons and protons which interact by effective forces .
all the information on the physical state of hot nuclei is then to be obtained from a careful study of their decay by emission of neutrons , protons , and @xmath0particles and their decay by fission .
the emission of photons will be neglected , as we consider energies far above the thresholds for particle emission .
the basis of our approach @xcite is a description of the particle emission as a purely statistical process , as given by the weikopf theory @xcite , and the nuclear fission as a transport process @xcite .
the importance of the non statistical aspects of the fission process in this context was recognized by grang and weidenmller @xcite .
our approach and also the ones of frbrich , abe and carjan @xcite are based on the premises of their work .
our efforts are dedicated to a stepwise improvement of this general theory in order to reach a quantitatively reliable description of the decay of the compound nuclei . in the present paper
, we give a more careful study of how the evaporation probabilities depend on the deformation of the nucleus , on its excitation and on its collection rotation . in the standard form of the evaporation theory @xcite ,
the deformation of the decaying nucleus enters only through the level densities of the initial and final nucleus .
we take into account that the transmission coefficient also depends on the deformation and , to a smaller degree , on the collective rotation of the nucleus .
this will be explained in chapter 2 .
our research is encouraged by the increasing amount of careful experimental studies of the evaporation of light particles from excited nuclei and of the concomitant decay by fission @xcite .
the results of our calculations are presented and compared with experimental work , especially of that of ref .
@xcite in chapter 3 .
finally , we summarize in chapter 4 our findings and point out the direction of our future work .
the fission process of hot and rotating nuclei should be described within a statistical model which takes into account the effect of energy dissipation due to the coupling to the internal degrees of freedom modelled by a friction force which generates diffusion .
the corresponding transport equation of the fokker - planck ( fpe ) type was originally proposed for nuclear fission by kramers in ref .
@xcite and later on adapted for heavy ions physics in ref .
@xcite - @xcite .
it is not easy to solve this equation exactly when the number of collective coordinates and conjugate momenta is larger than 2 .
an effective but approximate method of solving the fpe based on the moment expansion was used in @xcite .
unfortunately the probability distributions in the multidimensional space of collective coordinates is usually far from a gaussian form and the moment expansion method fails @xcite .
it is important to recall here that the set of differential equations which one obtains in the moment expansion method for the average coordinates and momenta is simply the set of classical equations of motion with friction as e.g. those used by bocki and co - workers @xcite . a more efficient way of solving the transport equations based on monte carlo method has been proposed in ref .
it was also shown in @xcite that the monte carlo calculation combined with the local moment expansion method transforms the fpe into a set of equations of motion containing random forces which is equivalent to a set of coupled langevin equations .
the langevin equation has already been used in @xcite in order to describe heavy - ion collisions and fission @xcite .
this idea is very fruitful and allows us to describe many phenomena occurring in heavy - ion physics and fission due to statistical fluctuations ( see e.g. the review article @xcite ) .
in chapter 2 we will present the physical basis of our model and explain how we couple the emission of light particles to the dynamical evolution of the system from the compound nucleus state to the scission point .
this dynamics is governed in our description by the langevin equation which we will present in sect .
2.1 , before explaining in sect .
2.2 how particle emission is incorporated into this description . in sect .
2.3 and 2.4 we deal with the initial conditions and the integration of the langevin equation .
+ we need to describe the time evolution of the nuclear system from an initial state which corresponds to a compound - nucleus at high excitation and angular velocity but usually created close to spherical symmetry to a final state in which a large amount of excitation and angular momentum will have been dissipated .
for such a description we use a single collective coordinate @xmath1 which measures the distance between the centers of mass of the two halves of the fissioning nucleus in units of the radius @xmath2 of the corresponding spherical nucleus . for the time being we restrict our description to symmetric fission .
the description of asymmetric fission would necessitate the introduction of an additional collective coordinate describing the mass asymmetry of the two emerging fission fragments .
if @xmath3 designates the conjugate momentum associated with the collective coordinate @xmath4 we obtain the following equations of motion describing the time evolution of the fissioning nucleus : @xmath5 @xmath6 + in these equations @xmath7 represents the collective mass , @xmath8 the potential and @xmath9 the friction coefficient @xcite .
the potential @xmath10 is calculated as the difference between the helmholtz free energies of the deformed and spherical nucleus .
the friction coefficient @xmath11 is calculated in the framework of the _ wall _ and _ window _ friction model @xcite .
the explicit expressions of these quantities , as obtained in the framework of the trentalange - koonin - sierk shape parametrization @xcite , have been presented and discussed in detail in ref .
the quantity @xmath12 designates the random langevin force which couples the collective dynamics to the intrinsic degrees of freedom .
this langevin force is chosen to be a gaussian random variable with zero mean value . in practice eqs .
( [ eq : lang1 ] ) - ( [ eq : lang2 ] ) are discretized by introducing a sufficiently small time step @xmath13 .
then the equations take the form @xmath14 @xmath15 + here @xmath16 where @xmath17 is a gaussian distributed random number with @xmath18 and @xmath19 and where the brackets represent ensemble averages . the diffusion coefficient @xmath20 appearing in eq .
( [ eq : ham2 ] ) is related to the friction coefficient @xmath9 through the einstein relation @xmath21 this relation holds in linear response theory as a high - temperature approximation of the dissipation - fluctuation theorem .
in the applications presented below the nuclear temperature will always be sufficiently high so that eq .
( [ eq : eins ] ) is approximately verified .
to speak at all about a nuclear temperature already supposes that the excitation of the nuclear system is shared to equal parts by all of the nucleonic degrees of freedom , so that statistical models apply . to simplify the description we consider the excited nucleus in a statistical description as member of a * grand - canonical ensemble * which can be characterized by a temperature @xmath22 .
for such a description to make sense , we need to assume that the time scale which governs the fission dynamics is sufficiently long compared to the one which determines the internal equilibration of the nuclear excitation among all the nucleons .
if this is the case , we can assume that the system can be considered as being continuously at equilibrium .
in this framework the temperature @xmath22 is simply a measure of the nuclear excitation energy @xmath23 and related to the the latter through the usual fermi gas relation @xmath24 where @xmath25 is the level density parameter of the considered nucleus at a nuclear deformation characterized by @xmath4 .
the excitation energy itself is determined by the conservation of the total energy as will be discussed in section 2.4 below .
+ the process of light particle emission from a compound nucleus is governed by the emission rate @xmath26 at which a particle of type @xmath27 ( neutrons , protons and @xmath0 particles are considered here ) is emitted at an energy in the range @xmath28 $ ] before the compound nucleus eventually undergoes fission .
several theoretical approaches have been proposed in order to describe the emission from a deformed , highly excited and rotating nucleus @xcite . the method which we present below is close in spirit to the one used in @xcite , but in our description the widths @xmath26 for particle emission will in addition depend on the deformation and angular momentum of the compound nucleus @xcite . according to weisskopf s conventional evaporation theory @xcite the partial decay rate @xmath29 for emission of a light particle of type @xmath27 with energy @xmath30 and orbital angular momentum @xmath31 from a compound nucleus with excitation energy @xmath23 rotating with an angular momentum @xmath32
can be written as @xmath33 where @xmath34 is the level density in the emitting nucleus .
the level density @xmath35 in the residual ( daughter ) nucleus with the excitation energy @xmath36 and the angular momentum @xmath37 is obtained in the same way .
both these quantities also depend on the mass and charge number and on the nuclear deformation .
the quantities @xmath38 and @xmath39 represent the moment of inertia and the level density parameter of the compound nucleus and @xmath40 the intrinsic spin of the emitted particle .
@xmath41 is the transmission coefficient for emitting a particle of type @xmath27 , with energy @xmath42 and angular momentum @xmath43 from the deformed compound nucleus .
the parameter @xmath44 in the argument list of @xmath45 stands for all quantities not explicitly mentioned here , such as the mass and charge number , the nuclear deformation , the direction in space in which the particle @xmath27 is emitted .
proceeding in this way would , however , leads to hardly tractable numerical problems , since in the langevin formalism which we endeavour here , we need to follow the dynamics of the fissioning nucleus plus evaporation for a very large number of trajectories ( of the order of 10@xmath46 ) .
we therefore use a simplified procedure which introduces a transmission coefficient @xmath47 obtained by a double averaging over the different emission directions and over the whole surface of the deformed compound nucleus .
a detailed description of how @xmath47 is calculated in the framework of the hill - wheeler approximation @xcite and how the averaging procedure is carried out is given in appendix a. with this averaged transmission coefficient @xmath48 the width @xmath26 for emission of a particle of type @xmath27 and energy @xmath30 reads : @xmath49 where @xmath50 is the angular momentum independent part of the density ( [ eq : rho ] ) and a similar relation holds for @xmath51 .
the effective transmission coefficient in ( [ eq : gamma2 ] ) is obtained by performing a summation over all allowed angular momenta of the emitted particle and those of the daughter nucleus : @xmath52 here @xmath53 is the maximal angular momentum available for the particle having the energy @xmath54 .
as already mentioned the emission width @xmath26 also depends on the mass and charge number a and z as well as on the deformation of the compound nucleus .
once the emission widths @xmath26 known , one can establish the emission algorithm which decides at each time step @xmath55 $ ] along each of the trajectories if a particle is emitted from the compound nucleus .
it is the value of the emission width @xmath26 which ultimately have to decide which of the light particles is emitted and at which energy .
since @xmath26 represents the rate at which a particle of type @xmath27 and energy @xmath30 is emitted , the total emission rate for a particle of given type , irrespectively of the energy at which it is emitted , is given by @xmath56 where we have replaced the upper integration limit by some large enough constant @xmath57 ( which will , a priori , be different for the different particles ) and at which the probabilities for particle emission will essentially have vanished .
@xmath58 is the number of energy bins of width @xmath59 in the interval @xmath60 $ ] .
the emission rate @xmath61 for emission of any kind of particles is the sum of the @xmath62 : @xmath63 first of all we need to decide whether a particle is emitted at all in the given time interval [ t , t + @xmath13 ] . the probability for emitting any particle
is given , for a small enough time step @xmath13 , by @xmath65 one then draws a random number @xmath66 in the interval @xmath67 $ ] . if @xmath68 a light particle is emitted .
if the time step @xmath13 is chosen sufficiently small , the probability for emitting a particle will be small . in this way we guarantee that in a time interval at most one particle is emitted and we avoid to consider the emission of more than one particle in each time interval . in the case that a particle is emitted one needs to decide next of which type this particle is . to this purpose one
draws a second random number @xmath69 in the interval @xmath67 $ ] and determines the localisation of @xmath69 with respect to the covering of this interval by the three bins @xmath70 , @xmath71 , and @xmath72 . depending on the bin in which @xmath69 is located , a neutron , or a proton or an @xmath0 particle
is emitted .
we still have to determine the energy with which this particle is emitted , and we accomplish this in the following way .
we introduce the quantity @xmath73 which represents the probability that a particle of type @xmath27 is emitted with an energy smaller than @xmath30 .
the quantity covers the interval @xmath67 $ ] . subdividing the interval @xmath67 $ ] in a certain number of equal bins , one decides with which energy the light particle
is emitted by drawing a third random number @xmath74 in the interval @xmath67 $ ] .
inverting the function in eq .
( [ eq : pi ] ) @xmath75 will fall in one energy bin @xmath76 { \label{eq : ealpha}}\ ] ] of the total interval @xmath77 $ ] and thus decide on the energy with which the particle is emitted .
+ the description of the fission process including particle emission starts , in principle , from an initial state corresponding to a spherical or deformed compound nucleus whose shape is characterized by the collective coordinate @xmath78 , the corresponding conjugate initial momentum @xmath79 , the excitation energy @xmath80 ( with corresponding temperature @xmath81 ) and the initial total angular momentum @xmath82 . in practice the excitation energy @xmath23 may be known approximately from the experiment .
this is , however , not the case for @xmath78 , @xmath79 and the angular momentum @xmath82 . in the calculations presented and discussed in sect .
4 we choose @xmath78 corresponding to a spherical nucleus and draw the conjugate momentum @xmath79 for each trajectory from a normalized gaussian distribution @xmath83\,\ , .
{ \label{eq : pini}}\ ] ] here @xmath84 is the collective inertia at the deformation @xmath78 .
it is in principle experimentally possible to determine the distribution of initial angular momenta for instance by measuring the @xmath11 multiplicities @xcite . as these distributions are rarely available , we chose @xmath82 to have a fixed value but we shall discuss below the dependence of the final results on the choice of @xmath82 .
+ once the initial conditions are fixed one can integrate the system of equations of motion ( 1 - 2 ) using their finite difference version ( 3 - 4 ) . at each time step @xmath85 $ ]
one draws a random number @xmath17 from a gaussian distribution which defines the fluctuating force in eq .
( 4 ) and thus generates a trajectory in the variable @xmath4 .
simultaneously one asks , as just explained , whether a light particle is emitted and decides , eventually , of its type @xmath27 and energy @xmath30 . at the beginning of the whole process
the system has a fixed total energy e@xmath86 which is the sum of the collective kinetic and potential energy @xmath87 and a collective rotation energy @xmath88 corresponding to a rigid deformed rotator of moment of inertia @xmath89 @xcite .
the excitation energy e@xmath90 at time @xmath91 and thereby the temperature of the nucleus is redetermined at each time step through the conservation of the total energy @xmath92 where the indices i or f refer to the initial ( time @xmath93 ) or final state ( time @xmath94 ) .
@xmath95 is the binding energy of the emitted particle ( which is zero unless an @xmath0 particle is emitted ) , @xmath30 is its kinetic energy after emission and @xmath96 is the recoil energy of the nucleus due to the emission process .
these three quantities enter the equation only when a light particle is emitted in the time interval @xmath85 $ ] .
since the recoil energy is in any case very small , it will be neglected .
additionally it is assumed that the deformation as well as the angular frequency of the nucleus is not changed due to the particle emission .
it is evident that each emission of a light particle carries away excitation energy and angular momentum and thereby increases the height of the fission barrier of the residual nucleus which , in turn , renders the fission event less and less probable . for each choice of the initial conditions one generates in this way a separate trajectory which is followed through in time . emitted particles and their energies are registered .
each trajectory can either lead to fission if it overcomes the fission barrier and continues on to the scission point , or can end up as a rather cold compound nucleus if too much energy has been lost to make the crossing of the fission barrier possible . sampling the total number of trajectories @xmath97 which have led to fission defines the fission cross section as @xmath98 with @xmath99 equal to the total number of trajectories .
let us finally mention that it is important to take into account the fact that the @xmath0 is a composite particle and that its emission therefore presupposes its existence in the nucleus prior to emission .
this is in principle quantified by the introduction of a preformation factor whose value can in principle be determined @xcite .
since this quantity is , however , very poorly known , we fix it in the following by multiplying the emission width @xmath100 by a factor @xmath101 ( @xmath102 ) . the choice of @xmath103 will be discussed in the following .
we study the decay of compound nuclei of several isotopes of @xmath104gd and of @xmath105yb at various excitation energies ranging from 150 mev to 300 mev . we selected these nuclei because a careful experimental investigation of their decays is available [ 7 ] and because they are known to exhibit large ground state deformations .
consequently , it is of special interest to investigate the influence of deformation on the emission of @xmath58 , @xmath3 and @xmath0 particles from excited states of these nuclei . the results concern the following physical aspects : * the dependence of the emission probabilities of neutron , protons , and @xmath0particles on the deformation , the excitation energy and the angular momentum of the emitting nucleus ( figs
. 1 to 9 ) . *
the dependence of the emission probabilities @xmath106 , eq .
( [ eq : pi ] ) , of a particle of given type @xmath27 on the kinetic energy @xmath42 ( figs . 10 to 15 ) . * the dependence , at different temperatures @xmath22 , of the number of fission events and light - particle multiplicities on the time t ( figs . 16 to 18 ) .
although these functions are not measurable , they are important for the physical understanding of the decay , especially the transient time phenomenon [ 4 ] and its dependence on the temperature . * the spectral distribution of the emitted particles ( figs .
19 to 21 ) . *
the dependence of the multiplicities , fission cross section and barriers on the initial angular momentum of the nucleus ( figs .
22 to 25 ) .
* the influence of the friction forces on the multiplicities of the emitted particles ( fig .
. we will now give a detailed description of these points . in figs .
13 , the emission width @xmath107 , eq .
( [ eq : gamma3 ] ) , for @xmath58 , @xmath3 , and @xmath0particles is shown as a function of the deformation @xmath108 of the emitting source .
the initial ensemble of decaying nuclei consists of @xmath109yb@xmath110 nuclei at an excitation energies of 50 mev , 150 mev and 250 mev respectively , and with a rotational angular momentum @xmath111 .
all the emission rates are seen to grow as a function of increasing deformation .
this trend can be easily understood as for increasing deformation the transmission occurs through a larger surface .
this effect has already been observed for all three particles @xcite .
we notice , however , that the emission width for @xmath0particles increases more steeply than the one for @xmath58 and @xmath3 for the two lower excitation energies @xmath112 = 50 mev ( fig . 1 ) and @xmath112 = 150 mev ( fig .
this is due to the fact that , as the nucleus is elongated , the barrier height for charged particles is reduced in the section of the surface which is farther away from the nuclear center and increased in the section which are closer to the center .
the section where the barrier is diminished is larger than the section where it is increased .
consequently , the emission rate for charged particles increases faster than the one for neutrons . in figs .
4 , 5 , and 6 the deformation dependences of the emission rates for @xmath58 , @xmath3 , and @xmath0 particles are shown for different values of the rotational angular momentum , varying between @xmath113 and 60 @xmath114 . for @xmath58 and @xmath3 the influence of the rotation on the emission rate is seen to be negligible , whereas the emission rate of @xmath0 particles grows by 10%20% as the rotational angular momentum increases from 0 @xmath114 to 60 @xmath114 .
it is clear that the centrifugal force , which helps to overcome the barrier , is largest for the @xmath0particle .
furthermore , the centrifugal effects matter the more the larger the barrier .
the barrier for the @xmath0particle is larger than the one for @xmath58 and @xmath3 .
let us point out that the rotational angular momentum of the nucleus has a very noticeable influence on the height of the fission barrier which decreases as a function of increasing angular momentum .
thus , at high angular momentum , nuclear fission can compete more effectively with evaporation . in figs .
79 , the deformation dependent emission width for @xmath58 , @xmath3 and @xmath0 is shown for two different isotopes of both @xmath104gd and @xmath105yb .
the results can be easily understood : for given proton number , the emission width @xmath115 for neutrons is the larger the larger the neutron number ( see fig .
7 ) . on the other hand , for given neutron number ,
the emission width @xmath116 for protons is the larger the larger the proton number ( see fig . 8) . at given proton number , the emission width @xmath117 , for @xmath0particles decreases with increasing neutron surplus whereas , at given neutron number , @xmath117 grows with increasing proton number ( see fig .
this is a simple consequence of the fact that the @xmath0particle contains an equal number of neutrons and protons .
an interesting quantity is the probability @xmath106 to emit a particle of type @xmath27 with an energy smaller than @xmath42 .
its definition in terms of the emission rate @xmath119 is given in eq .
( [ eq : pi ] ) .
the emission width @xmath62 depends also on the shape of the emitting nucleus , and so does the integrated probability @xmath106 . in figs . 10 to 12 ,
the probability @xmath106 is shown for the emission of neutrons , protons , and @xmath0particles . in each of the figures , @xmath106 is plotted separately for the case that the emitting nucleus has a spherical shape , the shape corresponding to the saddle and the scission point . in all cases
the fissioning nucleus is @xmath120yb at an initial excitation energy @xmath121 and an initial angular momentum @xmath122 .
all curves show a monotonous increase of @xmath123 as a function of the energy @xmath42 starting from a minimal energy @xmath42 which is zero for neutrons , but finite for protons and @xmath0particles due to the acceleration of the charged particles in the coulomb field of the residual nucleus . for neutron emission ,
the dependence of @xmath124 on the deformation of the emitting nucleus is very small , whereas for protons and @xmath0particles the curves @xmath106 are shifted to somewhat higher energies for the spherical source .
the physical interpretation is that the coulomb barrier is larger for the spherical nucleus than for the deformed nucleus and consequently the threshold for proton or @xmath0emission is at a higher energy than for neutrons .
we note that the scission shape for a nucleus like @xmath120yb consists essentially of two tangent , slightly deformed fragments of half the charge . in figs . 13 to 15
the probability @xmath106 is shown for neutrons , protons , and @xmath0particle emission for 3 different excitation energies of the initial compound nucleus @xmath109yb .
the shape of the emitting nucleus is chosen here as corresponding to the saddle point and its initial angular momentum is fixed at @xmath125 .
it is seen that the rise of the probability @xmath123 as a function of @xmath42 is the steeper the smaller the excitation energy .
such a behaviour can be easily understood since the range of energies of the emitted particles must rise with excitation energy .
we have also investigated the dependence of the functions @xmath106 on the neutron and proton numbers of the initial ensemble of compound nuclei , comparing the emission from the 4 nuclei @xmath126gd , @xmath127gd , @xmath128yb , and @xmath109yb , in each case for the same excitation energy ( 150 mev ) , the same angular momentum ( 40 @xmath114 ) and the same shape ( saddle point ) .
the results are indistinguishable for the emission of neutrons and only very slightly dependent on the nature of the emitting nucleus for the emission of protons and @xmath0particles .
the dependence of the number of decays of a given type on the time which elapses starting with the formation of the compound nucleus is unfortunately not measurable .
nevertheless , we think that it is interesting to exhibit this dependence for a few cases since it enables us to gain insight into the dynamical mechanism . in fig .
16 we show the number of fission events as a function of the time t on a logarithmic scale .
this result was obtained with the light - particle evaporation channels turned off .
the initial compound nucleus is @xmath109yb with an initial angular momentum @xmath125 .
the 3 curves in fig .
16 correspond to 3 different initial temperatures resulting in 3 different initial fission barrier heights @xmath129 .
it is seen that the transient time increases with decreasing excitation energy , as one expects .
it should be noted that the functions @xmath130 obtained for @xmath131 mev look quite similar to those for @xmath132 mev and are just shifted along the @xmath133 axis .
furthermore , if @xmath134 is the time where half of the final number @xmath135 of fission events have occurred @xmath136 we may approximate @xmath137 for times close to @xmath134 by a linear function of @xmath133 @xmath138 as one infers from fig .
16 , the dependence on the initial temperature is mainly contained in the quantity @xmath134 whereas the factor @xmath139 in ( [ eq:3.3 ] ) is almost the same for @xmath132 mev and @xmath131 mev .
the approximate time dependence ( [ eq:3.3 ] ) which is valid during part of the transient time interval is seen to be totally different from the one in the kramers regime which is valid in the cases when the fission barrier is much higher than the temperature of the fissioning nucleus . in fig .
17 we show the multiplicity ( number of emitted particles in coincidence with fission and per compound nucleus ) for the emission of neutrons , protons and @xmath0particles as a function of time for the decay of a compound nucleus @xmath109yb at an initial excitation energy of 293 mev and an initial angular momentum of 45 @xmath114 .
it is seen that the emission of @xmath3 and @xmath0particles ceases already after some 3@xmath14010@xmath141 sec whereas the number of emitted neutrons still increases .
again this is easily understood as a result of the different thresholds for charged and uncharged particles . in fig .
18 the fraction @xmath142 of nuclei undergoing fission is shown for the initial compound nucleus @xmath109yb as a function of time .
now , contrary to the results presented in fig .
16 , the emission of light particles is taken into account .
the initial excitation energy of the nucleus is @xmath143 mev and the initial angular momentum @xmath144 is assumed .
the well known kramers approximation holds whenever there is an approximately constant current across the fission barrier which implies a linear dependence of the number of fission events upon time .
it is seen from the figure that there is in fact no clearly distinguished section with a linear time - dependence . at best one
could replace the function by a straight line in the interval between 8 and 12@xmath14010@xmath141 sec . the `` transient time '' ,
i.e. the time needed for the build up of an approximately constant fission current is seen to be @xmath145 8@xmath14010@xmath141 sec .
this is the main part of the time available for fission events .
it is thus clear that the use of the kramers approximation or of the still simpler statistical transition state hypothesis for the whole time interval would lead to wrong results .
furthermore the very concept of a constant current across the fission barrier , which is the pre - requisite of the kramers approximation could loose its validity because of the existence of light particle emission which implies that the excitation energy and the angular momentum changes for those compound nuclei which emit a light particle prior to fission . in fig .
19 , we present the probability per energy unit of neutron , proton , and @xmath0particle emission as a function of the energy of the emitted particle . this figure is made on the basis of @xmath146 trajectories .
the solid curves show the spectral distribution in coincidence with fission , i.e. for particles emitted from nuclei which subsequently undergo fission .
the dashed curves represent the spectral distribution in anti coincidence with fission , i.e. for particles emitted from nuclei which subsequently end up as evaporation residues .
the initial compound nucleus is @xmath109yb at 293 mev excitation energy and an angular momentum of 52 @xmath114 .
the curves for different types of emitted particles are displaced against each other because charged particles gain energy in the coulomb field of the residual nucleus .
furthermore , the thresholds for emission of @xmath58 , @xmath3 , and @xmath0 are in general different . in addition , for @xmath0particles , the spectral distribution for particles emitted in coincidence with fission is slightly shifted towards smaller energies as compared to the distribution obtained when measured in anti - coincidence with fission .
this is probably due to the fact that charged particles are preferentially emitted from the pole tips around the long half axis .
the larger deformation then implies a smaller gain of kinetic energy from the repulsive coulomb field . in figs .
20 and 21 , the normalized yield for neutron emission is shown as a function of the nuclear deformation in the fission and evaporation channels .
the reader should notice that for nuclei which undergo fission , the emission of neutrons takes place , on the average , at a larger deformation than for the nuclei which end up as evaporation residues .
this is due to the fact that the distribution of the fissioning nuclei moves towards the saddle point in the deformation landscape
. this effect should give rise to an experimentally observed anisotropy in the angular distribution of prefission particles different from the one observed in the angular distribution of particles emitted by the evaporation residua .
the distribution for the case when the emission of protons and alpha particles is inhibited is shown in fig .
20 by the short dashed line .
it is seen that this distribution is very close to the one obtained when the emission of all three kinds of particles is allowed .
the initial angular momentum is different for the two figs .
20 and 21 .
it is seen that the maximum of the distribution of tne neutrons emitted by the evaporation residua is shifted with increasing angular momentum towards larger deformations .
the difference in the average deformation of the both distributions should be directly reflected in the difference in the anisotropy of the angular distributions of prefission neutrons and those emitted by the residua .
the multiplicity of prefission neutrons , protons and alpha particles is plotted in figs .
2223 as a function of the initial angular momentum @xmath32 .
the initial compound nucleus is @xmath120yb with the initial excitation energy e@xmath90=251 mev in fig .
22 and 293 mev in fig . 23 .
the corresponding fission rates are plotted in fig .
it is seen in figs .
22 23 that the neutron multiplicity decreases significantly with growing @xmath32 while that for protons and alphas is much less affected .
this is due to the fact that for large angular momenta the fission barriers @xmath129 become small ( see fig .
25 ) and it takes a shorter time to reach the scission configuration . consequently less neutrons are emitted on the average when the barrier is small .
alpha particles and protons are mostly emitted in the initial stage when the excitation energy of the nucleus is large , so that their multiplicities depend more weakly on the initial @xmath32 .
the effective fission rate ( including particle emission ) changes by 2 orders of magnitude ( see fig .
24 ) when one increases the value of the initial angular momentum by 12@xmath114 units up to its maximal value which corresponds to @xmath129=0 .
23 shows that this also holds true for the higher excitation energy of @xmath147293 mev .
this result indicates that a more precise knowledge of the initial spin distribution in compound nuclei is necessary in order to compare to the experiment .
one has to stress here that large @xmath32 values contribute most to the experimental value of the multiplicities as already seen in fig .
it therefore is certainly not correct to describe the experiment with a theoretical calculation using only the average value of the initial spin . as the fission barrier heights decrease strongly with increasing angular momentum as seen in fig .
25 , the higher l components which would lead to fission will have to be mocked up by an overestimation of the role of the fluctuating forces .
assuming an equal population of the initial angular momenta , which is , as just mentioned , not the best approximation , we have estimated the average of the @xmath58 , @xmath3 and @xmath0-multiplicities .
the resulting values are compared in table 1 with the experimental data taken from ref .
all parameters of the model are those given in sec .
2.1 and in ref .
we only have to assume the preformation factor @xmath148 ( see discussion in sec .
2.4 ) in order to reproduce the experimental number of alpha particles .
this goes into the right direction since our calculations show that @xmath0 particle emission is strongly enhanced by rotation and deformation effects .
the role of the friction forces in fission of high excited compound nuclei is widely known .
it was shown already in ref .
@xcite that the nuclear viscosity influence significantly the prescission neutron multiplicities .
this effect was discussed later in almost all papers dealing with the problem .
it was even assumed that the neutron multiplicities could be an indirect measure of the nuclear friction ( see e.g. ref .
@xcite ) .
the dependence of @xmath58 , @xmath3 and @xmath0-multiplicities on the magnitude of the friction coefficient connected with the elongation of the nucleus is plotted in fig . 26 for @xmath120yb at @xmath23=293
mev and @xmath32=50@xmath114 .
the strength of the friction coefficient is varied in the interval ( 0.02 , 2 ) in units of its value given by the wall formula @xcite .
the neutron multiplicity for small values of the friction coefficient is about 30% smaller than obtained with the wall friction .
as the transient time increases as a function of the friction , the increase of the multiplicity of neutrons emitted prior to fission is easly understood . again
the effect upon the emission of @xmath3 and @xmath0-particles is small , because these particles are emitted in a relatively short time interval after the formation of the compound nucleus which is in our case small compared to the transient time .
one has to notice that the results for very weak friction depend significantly both on the width of the initial distribution of the momentum conjugate with the fission coordinate ( eq .
( [ eq : pini ] ) ) and on the initial angular momentum . for larger values of the friction strength , close to @xmath149 ,
the influence of the initial distribution of the momentum @xmath79 on the neutron multiplicity is weak ( see also the discussion in ref .
our results demonstrate the importance of nuclear deformation on the evaporation of light particles from strongly excited nuclei .
this dependence on the deformation plays an important role also for the competition between fission and particle emission and might modify the limits which were determined for the nuclear friction force @xcite from the experimental data [ 6 ] on evaporation and fission .
the dependence of the evaporation width on the rotational angular momentum was found to be negligible for the emission of @xmath58 and @xmath3 and rather small for the emission of @xmath0particles .
let us , however , remind that the fission probability depends very sensitively on the angular momentum , as we have already emphasized in ref .
@xcite . due to this strong dependence of the fission probability on the initial angular momentum
, it becomes very important to get precise information on the angular momentum distribution of the initial ensemble of compound nuclei .
the outcome of the competition between light particle emission and fission may depend strongly on the initial angular momenta .
consequently , we have to try to obtain theoretical information on the angular momenta of the initial nuclei by treating the fission process dynamically @xcite .
agreement between calculated and measured emission probabilities for @xmath0particles can only be obtained , if an empirical preformation probability for @xmath0particles of about 0.2 is assumed .
one of the most important further improvements of the theory will be to evaluate this preformation factor within the temperature dependent thomas fermi approximation which underlies our theory @xcite .
another aspect which we intend to investigate is the angular dependence of emitted light particles when the angular momenta of the initial compound nuclei are aligned or the nuclei are polarized . in these cases it is conceivable that the angular dependence of the emitted particles shows a more pronounced dependence on the rotational angular momentum than the integrated yields we calculate in the present paper .
one might also hope that the angular distribution of emitted particles depends on the deformation of the source nuclei sufficiently sensitively so as to determine the deformation from such measurements .
of course , experimental data on the angular distribution of emitted neutrons , protons , and @xmath0particles from aligned rotating deformed nuclei would be of great interest for these studies .
one of us ( k.p . ) acknowledges gratefully the support during a four week stay in garching by the technical university munich and support by the in2p3 during a one month stay at c.r.n . in strasbourg .
j. b. , k. d. , and j. r. are very grateful for the kind hospitality extended to them at the umcs .
k. d. acknowledges financial support from the bmft and also from the daad for the travel funds to lublin .
10 e. strumberger , k. dietrich and k. pomorski , nucl . phys . *
a529 * ( 1991 ) 522 v. weisskopf , phys . rev .
* 52 * ( 1937 ) 295 h. kramers , physica * 7 * ( 1940 ) 284 p. grang , h.c .
pauli and h.a .
weidenmller , phys . lett .
* b88 * ( 1979 ) 9 ; z. phys .
* a296 * ( 1980 ) 107 p. frbrich , nucl . phys . * a545 * ( 1992 ) 87c g.r .
tillack , r. reif , a. schlke , p. frbrich , h.j .
krappe and h.g .
reusch , phys . lett .
* b296 * ( 1992 ) 296 y. abe , n. carjan , m. ohta , t. wada , proc .
in2p3riken symp . on heavy
ion collisions , obernai , 1990 , france d. hilscher and h. rossner , ann .
* 17 * ( 1992 ) 471 m. gonin , l. cooke , k. hagel , y. lou , j.b .
natowitz , r.p .
schmitt , s. shlomo , b. srivastava , w. turmel , h. ustunomiya , r. wada , g. nardelli , g. nebbia , g. viesti , r. zanon , b. fornal , g. prete , k. niita , s. hannuschke , p. gonthier and b. wilkins , phys . rev . * c42 * ( 1990 ) 2125 h. hofmann and p.j .
siemens , nucl . phys.*a257 * ( 1976 ) 165 , nucl .
a275 * ( 1977 ) 464 ch .
ng and h. hofmann , z. phys . *
a282 * ( 1977 ) 83 w. nrenberg , phys . lett .
* b52 * ( 1979 ) 289 h. hofmann in `` deep - inelastic and fusion reactions with heavy ions '' , berlin , 1979 , lecture notes in physics * 117 * , springer verlag , p. 64
+ f. scheuter and h. hofmann , nucl . phys . *
a394 * ( 1983 ) 477 j. bocki , h. feldmeier and w.j .
swiatecki , nucl.phys . *
a459 * ( 1986 ) 145 g. lamm and k. schulten , j. chem . phys .
* 78 * ( 1983 ) 2713 p. frbrich and s.y .
xu , nucl . phys . *
a477 * ( 1988 ) 143 y. abe , c. gregoire and h.delagrange , j. de phys . *
c4 * ( 1986 ) 47 j. bocki , j. randrup , w.j .
swiatecki , c.w .
tsang , ann .
* 105 * ( 1977 ) 427 j. bocki , y. boneh , j.r .
nix , j. randrup , m. robel , a.j .
sierk and w.j .
swiatecki , ann . phys .
* 113 * ( 1978 ) 330 s. trentalange , s.e .
koonin and a. sierk , phys .
* c22 * ( 1980 ) 1159 j. bartel , k. mahboub , j. richert and k. pomorski , z. phys . * a * , accepted for publication k. dietrich , k. pomorski and j. richert , z. phys . * a351 * ( 1995 ) 397 d.l .
hill , j.a .
wheeler , phys .
rev . * 89 * ( 1953 ) 1102 w. khn , a. ruckelshausen , r.d .
fisher , g. breibach , h.j .
hennrich , v. metag , r. novotny , r.v.f .
janssens , t.l .
khoo , d. habs , d. schwalm , b. haas and r.s .
simon , phys .
* 62 * ( 1989 ) 1103 k. bencheikh , p. quentin , j. bartel , nucl
a571 * ( 1994 ) 518 .
e. chabanat , et al .
, phys . lett . *
325b * ( 1994 ) 13 .
k. dietrich et al , in preparation .
m. blann , phys . rev .
* c21 * ( 1980 ) 1770 w. przystupa , k. pomorski , nucl
* a572 * ( 1994 ) 153 .
k. pomorski , w. przystupa , j. richert , acta phys . polon .
* b25 * ( 1994 ) 751 .
* appendix : evaluation of the transmission coefficients @xmath150 * + the explicit calculation of the transmission rates @xmath26 given by eq .
( [ eq : gamma2 ] ) requires the knowledge of the average transmission coefficient @xmath151 for the emission of light particles from the nucleus into the continuum . in principle
, a rigorous treatment would require the knowledge of @xmath152 which appears in eq .
( [ eq : gamma1 ] ) and should be calculated by averaging the transmission coefficients corresponding to the transmission of a particle emitted in a given direction from each point of the surface of the deformed , excited and rotating nucleus .
this can be done in principle but is hardly possible in practice where these calculations must be carried out at each time step and over a very large ensemble of trajectories .
we introduce the following simplified procedure which follows four steps : * we first calculate the transmission coefficients at three selected points ( 1,2,3 ) at which the main body - fixed axes ( x , y , z ) respectively cross the nuclear surface ( see fig . a1 ) .
the coordinate system is chosen such that the rotation axis coincides with the x - axis and the z - axis is the symmetry axis .
+ for all possible values of the angular momentum of the emitted particle @xmath153 where @xmath154 is the component of the vector @xmath155 parallel to the nuclear surface , and for each of the points ( i=1,2,3 ) @xmath156 is calculated using the hill - wheeler wkb expression @xcite @xmath157 \biggr)^{-1 } \;\ ; . \eqno(a2)\ ] ] here @xmath54 is the single - particle energy and the barrier height @xmath158 corresponds to the maximum of the potential @xmath159 in the direction @xmath160 perpendicular to the surface .
the potential in which the particle moves is given by : @xmath161 i.e. by the sum of nuclear @xcite , centrifugal and coulomb ( in the case of charged particles ) single - particle potentials .
the quantity @xmath162 appearing in ( a2 ) is given by @xmath163 * once @xmath164 is fixed at the three points ( i=1,2,3 ) one averages in a second step over all possible values of the x - component of @xmath43 which is the component along the rotation axis .
such a procedure seems to be reasonable once the angular momentum of the emitted particles is not measured experimentally .
+ through this average procedure one obtains @xmath165 * the transmission coefficients @xmath166 being determined at the three selected points ( i=1,2,3 ) one carries out in a third step an interpolation which allows to calculate the transmission coefficients to any point @xmath167 of the nuclear surface @xmath168 this interpolation of quadrupole type ensures that the transmission coefficient in the directions along the main axes are the same as those for emission in the opposite direction . *
finally one averages the coefficients over the whole nuclear surface : @xmath169 1 .
results of prescission neutron , proton and alpha multiplicity model calculations for @xmath120yb at excitation energies 251 mev and 293 mev compared with the experimental data taken from ref .
.the theoretical results are averaged over all initial angular momenta with a weight proportional to the corresponding fission rates . 1 . emission widths for neutrons ( n ) , protons ( p ) and alpha particles ( @xmath0 ) emitted from the hot , rotating compound nucleus @xmath120yb ( e@xmath90=50 mev , l=40@xmath114 ) as a function of the elongation .
2 . the same as in fig . 1 but for e@xmath90=150 mev .
3 . the same as in fig . 1 but for e@xmath90=250 mev .
emission widths for neutrons emitted from @xmath120yb at e@xmath90=150 mev and l=0 , 20 , 40 and 60@xmath114 as a function of the elongation .
the same as in fig . 4 but for protons . 6 .
the same as in fig . 4 but for alpha particles . 7 .
emission widths for neutrons emitted from different isotopes of gd and yb at e@xmath90=150 mev and l=40@xmath114 as a function of the elongation .
the same as in fig . 7
but for protons . 9 .
the same as in fig . 7
but for alpha particles . 10 .
probabilities to emit a neutron with energy smaller than @xmath170 from the fissioning nucleus @xmath120yb for three deformations corresponding to the spherical shape , the top of the fission barrier and the scission point . 11 .
the same as in fig .
10 but for protons . 12 .
the same as in fig .
10 but for alpha particles .
probability to emit a neutron with energy smaller than @xmath170 for three excitation energies ( e@xmath90=50 , 150 and 250 mev ) of the fissioning nucleus @xmath120yb .
the same as in fig .
13 but for protons . 15 .
the same as in fig .
13 but for alpha particles . 16 .
number of trajectories out of sample of 10.000 leading to fission as a function of time with particle emission turned off .
the calculation is done for three different temperatures ( t=3 , 4 and 5 mev ) of the fissioning nucleus @xmath120yb .
the heights of the corresponding fission barriers ( u@xmath171 ) are indicated in the figure .
multiplicity of the prefission neutrons ( n ) , protons ( p ) and alpha particles ( @xmath0 ) as a function of time .
ratio between the number of trajectories leading to fission and the total number of trajectories as a function of time for @xmath120yb .
the emission of the light particles is taken into account .
energy spectra of neutrons ( n ) , protons ( p ) and alpha particles emitted in coincidence with the fission events ( solid lines ) .
the dashed lines correspond to the spectra of the particles emitted from nuclei which end up as evaporation residua .
yield of neutron emission @xmath172 as a function of the deformation of the fissioning nucleus @xmath120yb with initial excitation energy e@xmath90=293 mev and initial angular momentum l=52@xmath114 .
the solid line corresponds to the trajectories which lead to fission while the dashed line to those leading to evaporation residua .
the dotted line describes the distribution when the emission of protons and alphas is turned off .
the same as in fig . 20
but for l=45@xmath114 .
multiplicity of prefission neutrons ( n ) , protons ( p ) and alpha particles ( @xmath0 ) as a function of angular momentum of the compound nucleus @xmath120yb with the initial excitation energy e@xmath90=251 mev .
the same as in fig .
22 but for the excitation energy e@xmath90=293 mev . 24 .
ratio between the number of trajectories leading to fission and the total number of trajectories as a function of the initial angular momentum for two different excitation energies e@xmath90=251 mev and 293 mev .
fission barrier heights as a function of the initial angular momentum for two different excitation energies e@xmath90=251 mev and 293 mev .
multiplicities of prescission neutrons ( n ) , protons ( p ) and alpha particles ( @xmath0 ) as a function of the strength of the friction force .
|
the dependence of the transmission coefficient on the deformation , the collective rotation and excitation energy of the compound nucleus emitting light particles is introduced in the framework of weikopf s evaporation theory . the competition between fission and particle evaporation is treated by a langevin equation for the fission variable coupled to the emission process .
detailed calculations are presented on the decay of different gd and yb isotopes at an excitation energy of about 250 mev .
these calculations demonstrate the importance of the effects of nuclear deformation and of the initial spin distribution on the evaporation .
|
grb 011121 was simultaneously detected by _ bepposax _
grbm and wfc ( piro 2001 ) , and the fluence in the 2 - 700 kev range corresponds to an isotropic energy of @xmath9ergs at the redshift of @xmath10 ( infante et al .
this burst was born in a stellar wind ( price et al .
2002 ; greiner et al . 2003 ) and a supernova bump was detected in the late optical afterglow ( bloom et al . 2002 ; garnavich et al .
its very early x - ray light curve , which has not been published until quite recently , is characterized by the presence of two flares ( prio 2005 , hereafter p05 ) . in the first one , which is also the strongest of the two ,
the observed flux @xmath11 rises and decays very steeply : @xmath12 for @xmath13 and @xmath14 for @xmath15 , where @xmath3 is the observer s time .
such a peculiar flare in the early x - ray lightcurve of grbs has not been predicted before .
p05 suggested that the x - ray flare represents the beginning of the afterglow . in this letter
we explore two alternative models which might give rise to very early x - ray flare in grb afterglows ( [ sec : xrb2 ] ) : a forward - reverse shock model ( [ sec : xrb21 ] ) and a late internal shock model " ( [ sec : xrb22 ] ) .
we compare the available data with the predictions of those models in [ sec : xrb3 ] and summarize our results in [ sec : xrb4 ] , with some discussion .
the external forward - reverse shock model has been widely accepted on interpreting the early ir / optical flashes of grb 990123 , grb 021211 and grb 041219a ( for observations , see : akerlof et al .
1999 ; fox et al .
2003 ; li et al . 2003 ; blake et al . 2005 . for theoretical modeling ,
see : sari & piran 1999 ; mszros & rees 1999 ; wei 2003 ; kumar & panaitescu 2003 ; fan , zhang & wei 2005 ) . synchrotron radiation from the reverse shock ( rs ) and
the forward shock ( fs ) usually peaks in the infrared - to - optical and ultraviolet - to - soft x - ray bands , respectively .
thus , the rs emission component is not dominant in the x - ray band .
the synchrotron self - compoton ( ssc ) scattering effect of the rs radiation has also been considered by different authors , but no strong x - ray emission is found to be expected ( wang , dai & lu 2001 ) except in some carefully balanced conditions ( kobayashi et al .
2005 ) . in most of previous works , the fractions of fs energy given to electrons , @xmath16 , and to magnetic field , @xmath17 , were assumed to be the same as the corresponding fractions in the rs . however , this may not necessary be the case . fan et al .
( 2002 ) performed a detailed fit to the optical flash of grb 990123 data and obtained @xmath18 and @xmath19 , where the superscripts `` r '' and `` f '' represent rs and fs , respectively .
similar results were obtained by zhang , kobayashi & mszros ( 2003 ) , kumar & panaitescu ( 2003 ) , panaitescu & kumar ( 2004 ) , mcmahon , kumar & panaitescu ( 2004 ) , and fan et al .
( 2005 ) . in this section
, we study the rs / fs emission in x - ray band by adopting different shock parameters .
we focus on the thin shell case ( i.e. , the rs is sub - relativistic , see kobayashi [ 2000 ] ) , in which the rs emission is well separated from the prompt @xmath0ray emission . * ism model . * in the thin shell case , the observer s time at which rs crosses the ejecta can be estimated by ( e.g. , fan et al . 2005 ) @xmath20 where @xmath21 is the isotropic energy of the outflow , @xmath22 is the typical number density of ism , and @xmath23 is the initial lorentz factor of the outflow . the convention @xmath24 has been adopted in cgs units throughout the text except for some special notations . in the standard afterglow model of a fireball interacting with a constant density medium ( e.g. , sari , piran & narayan 1998 ) , the cooling frequency @xmath25 , the typical synchrotron frequency @xmath26 , and the maximum spectral flux @xmath27 read : @xmath28 , @xmath29 @xmath30 , and @xmath31 , where @xmath32 $ ] , @xmath33 is the typical power - law distribution index of the electrons accelerated by fs , @xmath34/2 $ ] is the compton parameter , @xmath35 ( sari & esin 2001 ) , and @xmath36 is the luminosity distance for @xmath37 .
hereafter @xmath38 , and @xmath39 is in unit of days .
following zhang et al .
( 2003 ) , we take @xmath40 and @xmath41 . at @xmath1 ,
the rs emission satisfies [ see also fan et al .
( 2005 ) , note that a novel effect taken into account here is the inverse compton cooling of the electrons ] _ m,^r(t_)&=&r_b[r_e(_34,-1)]^2 _ m,^f(t_)/(_-1)^2 , + _ c,^r(t _ ) & & r_b^-3[(1+y^f ) /(1+y^r)]^2_c,^f(t
_ ) , + f _ , max^r(t _ ) & & _ b f _ , max^f(t _ ) . where @xmath42 is the lorentz factor of the shocked ejecta relative to the initial one
, @xmath43 is the bulk lorentz factor of the shocked ejecta at @xmath1 , @xmath44/2 $ ] is the compton parameter and @xmath45 .
if both @xmath46 and @xmath47 are below the observed frequency @xmath48 , the detected flux of rs and fs emission are f__^r(t_)f_,max^r(t _ ) [ _ m,^r(t_)]^(p-1)/2[_c,^r(t_)]^1/2 _ ^-p/2 , f__^f(t_)f_,max^f(t _ ) [ _ m,^f(t_)]^(p-1)/2[_c,^f(t_)]^1/2 _ ^-p/2 .
_ b^p-22r_e^p-1 ( _ 34,-1 _ -1)^p-1(1+y^f1+y^r ) .
[ eq : ratio1 ] taking @xmath49 ( such a choice is to match the observed slope of the x - ray flare @xmath50 , see p05 ) , @xmath51 , @xmath52 , @xmath53 , @xmath54 and @xmath55 , we have @xmath56 , @xmath57 , @xmath58 , @xmath59 , and @xmath60 , i.e. , we have a larger contrast @xmath61 when the inverse compton effect has been taken into account . with equation ( [ eq : ratio1 ] ) , we have @xmath62 , i.e. , _ in the x - ray band , the rs emission component is dominant_. for @xmath63 , the rs emission declines as @xmath64 because of the curvature effect ( e.g. , kumar & panaitescu 2000 , hereafter kp00 ) to describe the decline , which has been verified by the detailed numerical calculation ( fan , wei & wang 2004 ; see also fig .
[ fig : frx ] of this work ) . ] .
the fs emission declines as @xmath65 ( e.g. , sari et al . 1998 ) , so the x - ray flare lasts @xmath66^{\rm 4/(10-p)}t_{\times } \sim 300~{\rm s}$ ] .
taking @xmath10 , @xmath67 , and @xmath68 hz , we have @xmath69d_{\rm l,27.7}^{-2}$ ] .
the peak flux of the x - ray flare is @xmath70\sim 5\times 10^{-9}~{\rm ergs~cm^{-2}~s^{-1}}~[(1+z)/1.36]d_{\rm l,27.7}^{-2}$ ] , which is consistent with the observation of grb 011121 ( @xmath71 , see p05 ) .
however , the accompanying optical flash is very bright . with the typical parameters taken here ,
the v band flux is @xmath72 jy .
* wind model . *
grb 011121 was born in a stellar wind .
the best fit parameters are @xmath73 , @xmath74 , @xmath75 , @xmath76 , and @xmath77 ( p05 ) .
it is straightforward to show that with proper choice of @xmath78 and @xmath79 , at @xmath1 , the rs emission may be dominant in the soft x - ray band .
* numerical results . * following fan et al .
( 2005 ) , the fs - rs emission ( in the x - ray band ) has been calculated numerically . in the ism case ( see fig . [
fig : frx](a ) ) , the parameters are taken as @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , and the initial width of the outflow @xmath85 cm . in the wind case ( see fig . [
fig : frx](b ) ) , we take the best fit parameters presented in p05 . as shown in fig .
[ fig : frx](a ) , in the ism case , there comes an x - ray flare dominated by the rs emission only when both @xmath79 and @xmath78 are much larger than unity . in the wind case , with proper @xmath79 and @xmath78 , the rs emission may be dominant in the soft x - ray band , too .
but there is no flare expected since both the fs and rs emission components decrease continually even at very early time ( see also zou , wu & dai 2005 ) .
so the fs - rs scenario is unable to account for the x - ray flare detected in grb 011121 .
moreover , in the general framework of accounting for x - ray flares in grb afterglows , the fs - rs model is further disfavored for the fact that , even in an ism scenario with parameters suitable for a rs flare arising above the fs emission , the predicted temporal decay is too shallow compared to the steep decay of x - ray flares observed so far ( see fig .
[ fig : frx ] ) .
( 0,300 ) ( 0,0 ) ( 0,0 ) in the standard fireball model of grbs , the @xmath0ray emission is powered by internal shocks , whose duration depends on the active time of the central engine .
however , the variability of some grb afterglows implies that the activity of the grb central engine may last much longer than the duration of the prompt emission recorded by @xmath0ray monitors ( e.g. , dai & lu 1998 ; granot , nakar & piran 2003 ; ioka , kobayashi & zhang 2005 ) .
in addition , it has been proposed that the fe line observed in some grb x - ray afterglows could be attributed to a prolonged activity of the central engine ( rees & mszros 2000 ; gao & wei 2005 ) . a possible mechanism for the re - activity of the central engine could be as follows . during the accretion phase which powers the prompt @xmath0ray emission ,
a fraction of the material constituting the massive progenitor could possibly be pulled out ; the central engine could thus be restarted at late times by the fall back of part of this material onto the central collapsar remnant ( king et al . 2005 ) . here
we assume that the central engine restarts a few minutes after the prompt @xmath0ray emission , powering a new unsteady relativistic outflow .
we suppose that the lorentz factor of the ejected material can be highly variable , setting @xmath86 10 and @xmath87 100 , as the typical lorentz factors of the slow and fast shells , respectively .
the masses of the slow and fast shells are taken as @xmath88 . when an inner fast shell catches up with an outer slow shell at a radius @xmath89 ( where @xmath90 is the observed typical variability timescale of the x - ray flares ) ,
internal shocks are generated .
the lorentz factor of the merged shell is @xmath91 ( e.g. , piran 1999 ) , and the lorentz factor of the internal shocks can be estimated by @xmath92 .
we call this re - generated internal shocks as the late internal shocks " .
the internal shock model has been discussed by many authors ( e.g. , paczyski & xu 1994 ; rees & mszros 1994 ; daigne & mochkovitch 1998 , 2000 ; piran 1999 ; dai & lu 2002 ) . generally speaking , to calculate the synchrotron emission of internal shocks ,
the following parameters are involved : the outflow luminosity @xmath93 , @xmath90 , @xmath94 , @xmath95 , and of course the shock parameters @xmath16 and @xmath17 . for typical grbs , ( @xmath96 )
are taken to be @xmath97 , respectively ( e.g. , dai & lu 2002 ) . the time averaged isotropic luminosity ( 2 - 700 kev ) of the x - ray re - bursting of grb 011121 is @xmath98 ( p05 ) .
so we normalize our expression by taking @xmath99 , where @xmath100 is the efficiency factor of the x - ray flare .
we take @xmath101 , as found in typical grbs / xrfs ( lloyd - ronning & zhang 2004 ) . such small @xmath93 ( comparing with the grb case ) implies that the fallback accretion rate is just @xmath102 times that of the prompt accretion , if the efficiency factor of converting the accretion energy into the kinetic energy of the outflow is nearly a constant ( macfadyen , woosley & heger 2001 ) .
the @xmath90 measured in x - ray flares is significantly longer than that measured in the intrinsic hard burst ( burrows et al .
2005 ) . in this letter
, we take @xmath103 10 s. the spectra of the x - ray flares detected so far are all nonthermal .
the optical thin condition implies a lower limit on the bulk lorentz factor of the merged shell ( e.g. , rees & mszros 1994 ) > 11 l_m,49.7 ^ 1/5[1.36/(1+z)]^-1/5t_,1 ^ -1/5 , the typical radius of the late internal shock is @xmath104^{3/5}\delta t_{\oplus,1}^{3/5}$ ] . as we will show in the following section , with @xmath105 , @xmath106 s , @xmath107 , @xmath108 , @xmath109 , and @xmath110 , most of the internal shocks energy is emitted in the soft x - ray band and the predicted flux matches the observation of the x - ray flare in grb 011121 .
alternatively , as shown in barraud et al .
( 2005 ) , with @xmath111 ( correspondingly , @xmath112 and @xmath113 ) , @xmath103 a few seconds , and @xmath114 hundreds , x - ray dominated luminosity is expected , too ( please note that their calculation of the internal shocks emission is somewhat different from ours ; see barraud et al .
2005 for detail ) .
therefore , we believe that with proper parameters , x - ray flares do appear in the late internal shocks scenario .
following dai & lu ( 2002 ) , the comoving number density of the unshocked outflow is estimated by @xmath115 , where @xmath116 is the rest mass of proton .
the thermal energy density of the shocked material is calculated by @xmath117 ( blandford & mckee 1977 ) .
the intensity of the generated magnetic field is estimated by @xmath118^{1/2 } l_{\rm m,49.7}^{1/2}\gamma_{1.5}^{-1}r_{\rm int,14.5}^{-1}$ ] . as usual
, we assume that in the shock front , the accelerated electrons distribute as @xmath119 , where @xmath120/ [ ( p-1)m_{\rm e}]$ ] is the minimum lorentz factor of the shocked electrons ( sari et al .
1998 ) , and @xmath121 is the rest mass of electron . in this section , we take p=2.5 . the observed typical frequency of the synchrotron radiation reads _ m , & = & _ e , m^2q_e b/[2(1+z)m_ec ] + & & 2.710 ^ 16 hz _
e,-0.3 ^ 2_b,-1 ^ 1/2(_sh-1)^5/2(_sh/2)^1/2 + & & l_m,49.7 ^ 1/2_1.5 ^ -2t_,1 ^ -1 , [ eq : nu_m ] i.e. , most of the shock energy is emitted in the soft x - ray band , where @xmath122 is the charge of electron .
the cooling lorentz factor is estimated by ( e.g. , sari et al . 1998 ) @xmath123 , and the corresponding cooling frequency reads _
c , & = & _ e , c^2 q_e b/[2(1+z)m_ec ] + & & 10 ^ 10 hz [ 1.36/(1+z)]^2 _ b,-1 ^ -3/2_1.5 ^ 8 + & & [ _ sh(_sh-1)/2]^-3/2 l_m,49.7 ^ -3/2t_,1 .
[ eq : nu_c ] the synchrotron self - absorption frequency is estimated by ( li & song 2004 ) _
a,10 ^ 15 hz [ 1.36/(1+z)]^3/7l_m,49.7 ^ 2/7_1.5 ^ -5/7t_,1 ^ -4/7b_3 ^ 1/7 .
the maximum spectral flux of the synchrotron radiation is ( e.g. , wijers & galama 1999 ) @xmath124 , where @xmath125=8\times 10^{51}~[1.36/(1+z)]l_{\rm m,49.7}\gamma_{1.5}^{-1 } \delta t_{\oplus,1}$ ] is the number of electrons involved in the emission .
@xmath126 is a function of @xmath127 , for @xmath73 , @xmath128 ( wijers & galama 1999 ) . for @xmath129 ,
the predicted flux is ( e.g. , sari et al . 1998 ) f__x & = & f_max(_m , /_c , ) ^-1/2(_x/_m , ) ^-p/2 + & & 2.5 mjy [ _ x/(2.4210 ^ 17hz)]^-p/2_e,-0.3^p-1_b,-1^(p-2)/4 + & & ( _ sh/2)^(p-2)/4 ( _ sh-1)^(5p-6)/4l_m,49.7^(p+2)/4 + & & _ 1.5 ^ 2-pt_,1^(2-p)/2 d_l,27.7 ^ -2 .
[ eq : flux ] taking @xmath130 and @xmath131 hz , with equation ( [ eq : flux ] ) we have @xmath132 mjy , which matches the observation of grb 011121 ( @xmath133 mjy ) .
the v band flux can be estimated as ( @xmath134 ) @xmath135 mjy . what happens after the late internal shocks " ?
surely , the refreshed relativistic outflow will catch up with the initial outflow when the latter has swept a large amount of material and got decelerated . that energy injection would give rise to a flattening ( e.g. , rees & mszros 1998 ) or re - brightening signature ( e.g. , panaitescu , mszros & rees 1998 ; kumar & piran 2000 ; zhang & mszros 2002 ) , which could potentially account for the late re - brightening of xrf 050406 , the late x - ray re - bursting detected in grb 050502b and
the second / weaker x - ray bump observed in grb 011121 .
however , the detailed lightcurve modeling is beyond the scope of this letter . for observer s frequencies above the cooling frequency
, the curvature effect dominates the temporal behavior of the observed flux after the central engine shuts down at @xmath6 ( pk00 ) .
the flux declines as @xmath136^{\rm -(2+p_i/2)}$ ] ( pk00 ) , where @xmath137 represents the @xmath137th pulse , @xmath138 and @xmath139 are the ejection time and the variability timescale of the @xmath137th pulse , respectively .
such a decline is much steeper than @xmath140 for @xmath141 .
for example , as shown in fig .
[ fig : illustration ] , the decline of the flare is dominated by the curvature effect of the last long pulse with @xmath142 .
a crude power - law fit to the decline yields @xmath143 , which is steep enough to match the sharpest decline detected so far . in reality
, the central engine does not turn off abruptly .
the dimmer and dimmer emission powered by the weaker and weaker late internal shocks " may dominate over the curvature effect of the early pulses , resulting in a shallower decay .
( 0,200 ) ( 0,0 )
early x - ray flares have been well detected in grb 011121 , xrf 050406 , grb 050502b and grb 050730 ( p05 ; burrows et al .
2005 ; staring et al .
the rise and fall of the first flare ( also the dominant one ) in grb 011121 are both very steep .
similar temporal behavior is evident in other events .
the sharp decline of these flares imposes a robust constraint on the model , as shown below .
the x - ray flare detected in grb 011121 appears at @xmath144 s and peaks at @xmath145 s. the burst is believed to be born in a weak stellar wind . as shown in fig .
[ fig : frx](b ) , no flare is expected in the fs - rs model .
the fs - rs shock model is further disfavored by its shallow decline . in the late internal shock model ,
the decline of the flare can be steep enough to account for the observation ( see fig .
[ fig : illustration ] for illustration ) .
moreover , as shown in
[ sec : radition ] , with proper parameters the observed flux can be well reproduced .
so the late internal shock model " is favored .
we would like to point out that _ the fall of the x - ray flare detected in grb 011121 is still attributed to the late internal shocks rather than the curvature effect_. the reason is as follows . since @xmath146 s , the resulted decline @xmath147^{-3.15}$ ] is much steeper than the observation @xmath148^{-1.4}$ ] ( see fig .
7 of p05 ) if after @xmath149 there are no internal shocks any more .
the x - ray flare detected in xrf 050406 peaks at @xmath150 s and declines as @xmath151 .
the x - ray flare detected in grb 050502b peaks at @xmath152 s and declines as @xmath153 . in the x - ray afterglow lightcurve of grb 050730
, there are three x - ray flares ( ranging from 200 s to 800 s after the trigger of the grb ) .
a crude fit to the decline of these three flares results in @xmath154 or steeper .
obviously , the fs - rs scenario is ruled out by the steep observed decays and the late internal shock model " is favored . for the x - ray flare detected in grb 050502b ,
the late internal shock model interpretation is further supported by the sharp spike detected in [email protected] kev band ( burrows et al . 2005 ) .
in this work , we have explored two possible models which might give rise to x - ray flares in grb afterglows .
one is the external forward - reverse shock model ( the ism case ) , in which the shock parameters of forward / reverse shocks are taken to be quite different . _
the other is the late internal shock model " , which requires that a refreshed unsteady relativistic outflow is generated after the prompt @xmath0ray emission _ ( see ramirez - ruiz , merloni & rees [ 2001 ] for alternative scenarios ) , _ perhaps due to the fallback accretion onto the central collapsar remnant_. the refreshed outflow may be characterized by a low outflow luminosity ( @xmath155 ) , a small bulk lorentz factor ( @xmath156 ) , and a long variability timescale ( @xmath157 s ) . in the external forward - reverse shock model , after the peak of the reverse shock emission ( @xmath158 ) , the flux can not decline more sharply than @xmath159 ( see fig . [
fig : frx ] for illustration ) . in the late internal shock model "
, the decline can be much steeper than @xmath160 if the central engine shuts down after @xmath161 and the longest variability timescale of the x - ray flare is much shorter than @xmath161 ( see fig .
[ fig : illustration ] for illustration ) . for the x - ray flares detected in grb 011121 ,
xrf 050406 , grb 050502b and grb 050730 , the external forward - reverse shock model is ruled out directly by its shallow temporal decay .
for the same reason , other possible external models ( i.e. , the model related to the external forward shock ) , including the density jump model , the two - components jet model , the patch jet model as well as the energy injection model are ruled out too ( zhang et al .
thus , the late internal shock model " is found to be favored . in this model
, the optical emission may be suppressed due to strong synchrotron - self - absorption .
but in the ultraviolet band , the radiation could be quite strong .
large amount of neutral gas would be ionized , as detected in grb 050502b and grb 050730 ( burrows et al . 2005 ; starling et al .
2005 ) .
_ very early x - ray flares are well detected both in long grbs and in xrfs , which strengthens the correlation of these two phenomena _
, though the nature of xrfs is still unclear ( barraud et al . 2005 and the references therein ) .
finally , we suggest that the early x - ray light curve of some grbs may be a superposition of the emission powered by the long activity of the central engine and the emission of the external forward shock . as a consequence , the x - ray temporal behavior may be quite different from that of the long wavelength emission ( uv / optical ones ) .
this prediction can be tested by the uvot and xrt on board _ swift _ observatory directly in the near future .
we thank bing zhang and e. w. liang for informing us piro et al.s paper on grb 011121 at the end of feb , and g. f. jiang , l. j. gou , z. li and h. t. ma for kind help .
we also appreciate the referees for their helpful comments and the third referee for her / his great help . this work is supported by the national natural science foundation ( grants 10225314 and 10233010 ) of china , and the national 973 project on fundamental researches of china ( nkbrsf g19990754 ) .
99 akerlof c. et al .
1999 , nature , 398 , 400 barraud c. , daigne f. , mochkovitch r. , atteia j. l. 2005 , a&a , in press ( astro - ph/0507173 ) blandford , r. d. , & mckee , c. f. 1977 , mnras , 180 , 343 blake c. h. et al .
2005 , nature , 435 , 181 bloom j. s. , et al .
2002 , apj , 572 , l45 burrows , d. n. , et al .
2005 , science , in press ( astro - ph/0506130 ) dai z. g. , lu t. , 1998 , a&a , 333 , l87 dai z. g. , lu t. , 2002 , apj , 580 , 1013 daigne f. , mochkovitch r. 1998 , mnras , 296 , 275 daigne f. , mochkovitch r. 2000 , a&a , 358 , 1157 fan y. z. , dai z. g. , huang y. f. , lu t. , 2002 , chin .
j. astron .
astrophys . 2 , 449 ( astro - ph/0306024
) fan y. z. , wei d. m. , wang c. f. , 2004 , a&a , 424 , 477 fan y. z. , zhang b. , wei d. m. , 2005 , apj , 628 , l25 fox d. et al .
2003 , apj , 586 , l5 gao w. h. , wei d. m. , 2005 , apj , 628 , 853 garnavich p. m. , et al .
2003 , apj , 582 , 924 granot j. , nakar e. , piran t. , 2003 , nature , 426 , 138 greiner j. , et al .
2003 , apj , 599 , 1223 infante l. , garnavich p. m. , stanek k. z. , wyrzykowski l. , 2001 , gcn circ .
, 1152 ioka k. , kobayashi s. , zhang b. , 2005 , apj , in press ( astro - ph/0409376 ) king a. , obrien p. t , goad m. r. , osborne j. , olsson e. , page k. 2005 , apjl , in press ( astro - ph/0508126 ) kobayashi s. 2000 , apj , 545 , 807 kobayashi s. , zhang b. , mszros p. , burrows w. , 2005 , apjl , submitted ( astro - ph/0506157 ) kumar p. , panaitescu a. , 2003 , mnras , 346 , 905 kumar p. , panaitescu a. , 2000 , apj , 541 , l51 ( kp00 ) kumar p. , piran t. , 2000 , apj , 532 , 286 li z. , song l. m. 2004 , apj , 608 , l17 li w. d. , filippenko a. v. , chornock r. , jha s. , 2003 , apj , 586 , l9 lloyd - ronning , n. m. , & zhang , b. 2004 , apj , 601 , 371 macfadyen a. i. , woosley s. e. , herger a. 2001 , apj , 550 , 410 mcmahon e. , kumar p. , panaitescu a. , 2004 , mnras , 354 , 915 mszros p. , rees m. j. , 1999 , mnras , 306 , l39 paczyski b. , xu g. h. , 1994 , apj , 427 , 708 panaitescu a. , kumar p. , 2004 ,
mnras , 353 , 511 panaitescu a. , mszros p. , rees m. j. , 1998 , apj , 503 , 314 piran t. , 1999 , phys .
, 314 , 575 piro l. , 2001 , gcn circ .
1147 piro l. et al . , 2005 ,
apj , 623 , 314 ( p05 ) price p. a. , et al .
2002 , apj , 572 , l51 ramirez - ruiz e. , merloni a. , rees m. j. , 2001 , mnras , 324 , 1147 rees m. j. , mszros p. , 1994
, apj , 430 , l93 rees m. j. , mszros p. , 1998 ,
apj , 496 , l1 rees m. j. , mszros p. , 2000 , apj , 545 , l73 sari r. , esin a. a. 2001 , apj , 548 , 787 sari r. , piran t. 1999 , apj , 517 , l109 sari r. , piran t. , narayan r. 1998 , apj , 497 , l17 starling r. l. c. , et al .
2005 , a&a , submitted ( astro - ph/0508237 ) wang x. y. , dai z. g. , lu t. , 2001 , apj , 556 , 1010 wei d. m. 2003 , a&a , 402 , l9 wijers r. a. m. j. , & galama t. j. 1999 , apj , 523 , 177 zhang b. , fan y. z. , dyks j. , kobayashi s. , mszros p. , burrows d. n. , nousek j. a. , gehrels n. 2005 , apj , submitted ( astro - ph/0508321 ) zhang b. , kobayshi s. , mszros p. , 2003
, apj , 595 , 950 zhang b. , mszros p. , 2002
, apj , 566 , 712 zou y. c. , wu x. f. , dai z. g. , 2005 , mnras , in press ( astro - ph/0508602 )
assuming a shell is ejected at @xmath162 and moves with a constant lorentz factor @xmath94 ( the corresponding velocity is @xmath163 , in unit of the speed of light ) , the electrons are shock - accelerated at @xmath164 and cools rapidly , where @xmath165 is the radius after which there is no newly relativistic electrons injected . for @xmath166 , the radiation nearly cuts off as long as the observer frequency @xmath48 is above the cooling frequency of the electrons .
for @xmath167 , the flux received from the shell is given by @xmath168 where @xmath169 satisfies @xmath170 $ ] , @xmath171 is the half opening angle of the shell , @xmath172 , @xmath173 is the specific spectrum of the radiation in unit of solid angle . on the equal arriving time surface "
@xmath174 $ ] ( rees 1966 , nature , 211 , 468 ) , the magnetic field , the typical emission frequency of the electrons , and the number of electrons involved in the radiation are all functions of @xmath175 , that s why we take into account the term @xmath176 in calculating @xmath177 . equation ( [ eq : curv1 ] ) yields @xmath178 on the other hand , @xmath179 . for @xmath180 , we have @xmath181^{\rm -(2+\beta)}\propto ( { t_\oplus - t_{\rm eje}\over t_0-t_{\rm eje}})^{\rm -(2+\beta ) } , \label{eq : curv3}\ ] ] which coincides with that presented in kp00 but derived in a different way .
|
we explore two possible models which might give rise to bright x - ray flares in grbs afterglows .
one is an external forward - reverse shock model , in which the shock parameters of forward / reverse shocks are taken to be quite different .
the other is a so called late internal shock model " , which requires a refreshed unsteady relativistic outflow generated after the prompt @xmath0ray emission . in the forward - reverse shock model , after the time @xmath1
at which the rs crosses the ejecta , the flux declines more slowly than @xmath2 , where @xmath3 denotes the observer s time and @xmath4 is the spectral index of the x - ray emission . in the late internal shock model " , decaying slopes much steeper than @xmath5 are possible
if the central engine shuts down after @xmath6 and the observed variability timescale of the x - ray flare is much shorter than @xmath7 .
the sharp decline of the x - ray flares detected in grb 011121 , xrf 050406 , grb 050502b , and grb 050730 rules out the external forward - reverse shock model directly and favors the late internal shock model " .
these x - ray flares could thus hint that the central engine operates again and a new unsteady relativistic outflow is generated just a few minutes after the intrinsic hard burst .
gamma rays : bursts@xmath8ism : jets and outflows radiation mechanisms : nonthermal@xmath8x - rays : general
|
it is a pleasure to thank masaaki kuroda , bernd surrow and mikhael tentyukov for a fruitful collaboration on the subject matter of the present talk .
zeus 96/97 : s. chekanov et al .
, zeus collab .
, desy-01 - 014 , submitted to eur . phys .
j. c ; m. derrick et al .
, z. phys .
c72 , 399 ( 1996 ) ; zeus svtx 95 : j. breitweg et al . ,
zeus collab .
j c7 , 609 ( 1999 ) ; zeus bpc 95f : j. breitweg et al . , zeus collab .
b407 , 432 ( 1997 ) ; zeus bpt 97 : j. breitweg et al .
, zeus collab .
b487 , 53 ( 2000 ) .
h1 svtx 95 : c. adloff et al . , h1 collab .
b497 , 3 ( 1997 ) ; h1 96/97 : c. adloff et al . , h1 collab . , accepted by euro .
phys . j. c ; h1 97 : c. adloff et al . , h1 collab . , euro
j. c13 , 609 ( 2000 ) .
d. schildknecht , presented at diffraction2000 , cetraro ( italy ) september 2000 , nucl .
b ( proc . suppl . ) 99 , 121 ( 2001 ) ; + d. schildknecht , b. surrow , m. tentyukov , phys . lett .
b499 , 116 ( 2001 ) ; + g. cvetic , d. schildknecht , b. surrow , m. tentyukov , euro .
j. c20 , 77 ( 2001 ) .
|
the empirical scaling law , wherein the total photoabsorption cross section depends on the single variable @xmath0 , provides empirical evidence for saturation in the sense of @xmath1 for @xmath2 at fixed @xmath3 .
the total photoabsorption cross section is related to elastic diffraction in terms of a sum rule .
the excess of diffractive production over the elastic component is due to inelastic diffraction that contains the production of hadronic states of higher spins .
motivated by the diffractive mass spectrum , the generalized vector dominance / color dipole picture ( gvd / cdp ) is extended to successfully describe the dis data in the full region of @xmath4 , all @xmath5 , where the diffractive two - gluon - exchange mechanism dominates . in the present talk ,
i wish to concentrate on the relation between the total photoabsorption cross section , @xmath6 , at low @xmath7 and diffractive production , @xmath8 @xcite .
the experimental data @xcite on @xmath6 at @xmath4 and all @xmath5 , including photoproduction @xmath9 , lie on a single curve @xcite , _ ^
* p ( w^2 , q^2 ) = _ ^*p ( ( w^2 , q^2 ) ) , [ ( 1 ) ] if plotted against the low - x scaling variable ( w^2 , q^2 ) = , [ ( 2 ) ] where @xmath10 is a slowly increasing function of @xmath11 and @xmath12 .
compare fig . 1 for a plot of @xmath6 against @xmath13 .
the function @xmath14 may be represented , alternatively , by a power law or by a logarithm , ^2 ( w^2 ) = \ { .
[ ( 3 ) ] as a function of @xmath15 compared with the predictions from the gvd / cdp.,width=245 ] we refer to refs.@xcite for the numerical values of the fit parameters in @xmath14 . the empirical model - independent finding ( [ ( 1 ) ] )
is interpreted in the generalized vector dominance / color dipole picture ( gvd / cdp ) @xcite that rests on the generic structure of the two - gluon - exchange virtual - photon - forward - compton - scattering amplitude .
evaluation of this amplitude in the @xmath16 limit and transition to transverse position space implies @xcite @xmath17 where the fourier representation of the color - dipole cross section , @xmath18 contains `` color transparency '' in the limit of @xmath19 , as well as hadronic unitarity , provided ^ ( ) d l^ 2_(l^ 2 _ , z , w^2 ) [ ( 6 ) ] has decent high - energy behavior .
the average or effective gluon transverse momentum , @xmath20 , in ( [ ( 5 ) ] ) is given by l^
2__w^2 , z .
[ ( 7 ) ] it is a characteristic feature of the @xmath21 limit of the two - gluon - exchange amplitude that the representation ( [ ( 4 ) ] ) factorizes into the product of the photon wave function , @xmath22 , that describes the photon coupling to the @xmath23 state and its propagation , and the color - dipole cross section , @xmath24 , that describes the forward scattering of the color dipole from the proton .
the scattering is `` diagonal '' in the variables @xmath25 , since these variables remain fixed during the scattering process . the empirical scaling law ( [ ( 1 ) ] ) is embodied in the representation ( [ ( 4 ) ] ) by requiring the dipole cross section ( [ ( 5 ) ] ) to depend on the product @xmath26 .
this implies that @xmath27 be proportional to @xmath14 . in the gvd / cdp
, we approximate the distribution in the gluon transverse momentum , @xmath28 , in ( [ ( 5 ) ] ) by a @xmath29-function situated at the effective gluon transverse momentum , @xmath20 , @xmath30 with ( [ ( 5 ) ] ) and ( [ ( 8) ] ) , and the fourier representation of the wave function inserted , the expression for the cross section ( [ ( 4 ) ] ) may be evaluated analytically in momentum space @xcite .
we only note the approximate final expression @xmath31 and refer to ref.@xcite for details . according to ( [ ( 9 ) ] ) , at any fixed value of @xmath3 , for sufficiently large @xmath32 , a soft , logarithmic energy dependence is reached for @xmath33 .
the gvd / cdp that rests on the generic structure of the two - gluon exchange from qcd , and contains hadronic unitarity and scaling in @xmath13 , leads to the important conclusion that _ w^2 = 1 . [ ( 10 ) ] the behavior ( [ ( 9 ) ] ) may be called `` saturation '' . since the low x ( hera ) data , according to fig . 1 ,
show evidence for the behavior ( [ ( 9 ) ] ) that implies ( [ ( 10 ) ] ) , we may indeed conclude that hera yields evidence for `` saturation '' .
needless to stress , future tests of scaling in @xmath13 , by increasing @xmath32 as much as possible , are clearly desirable to provide further evidence for the validity of the remarkable conclusion ( [ ( 10 ) ] ) that puts virtual and real photoproduction on equal footing at any fixed @xmath3 in the limit of infinite energy .
we turn to diffractive production .
the two - gluon - exchange generic structure for @xmath21 implies @xcite @xmath34 note the close analogy of ( [ ( 11 ) ] ) to the simple @xmath35 dominance formula for photoproduction @xcite upon transition to the momentum - space representation in ( [ ( 11 ) ] ) and after integration over all variables with the exception of the mass @xmath36 of the outgoing state @xmath37 , one obtains the mass spectrum , @xmath38 for forward production that depends on @xmath39 and @xmath40 .
a comparison of this mass spectrum with the integrand of the total cross section in ( [ ( 4 ) ] ) ( obtained upon transition to momentum space and appropriate integration with the exception of one final integration over @xmath40 ) , allows one to rewrite ( [ ( 4 ) ] ) as a sum rule that reads @xcite @xmath41 .\nonumber\end{aligned}\ ] ] the sum rule represents the total photoabsorption cross section in terms of diffractive forward production .
it is amusing to note that ( [ ( 13 ) ] ) is the virtual - photon analogue of the photoproduction sum rule @xcite = _
v= ^0 , , , ... [ ( 14 ) ] based on @xmath42 dominance .
note , however , that ( [ ( 13 ) ] ) is a strict consequence of the generic two - gluon exchange structure evaluated in the @xmath21 limit that forms the basis of the gvd / cdp ) is also obtained from gvd arguments by themselves @xcite ] .
it is evident , even though apparently always ignored , that the diffractive production cross section ( [ ( 11 ) ] ) describes elastic and only elastic diffraction , where `` elastic '' is meant to denote diffractive production of hadronic states @xmath37 that carry photon quantum numbers .
otherwise , the color dipole cross section under the integral in ( [ ( 11 ) ] ) could never be identical to the one in ( [ ( 4 ) ] ) , and ( [ ( 13 ) ] ) could never follow from ( [ ( 4 ) ] ) and ( [ ( 11 ) ] ) .
`` inelastic '' diffraction , namely diffractive production of states with spins different from the projectile spin , subject to the restriction of natural parity exchange , is a well - known phenomenon in hadron physics @xcite .
evidence for inelastic diffraction in dis is provided by the decrease @xcite of the average thrust angle ( `` alignment '' ) with increasing mass of the produced state @xmath37 .
this observation implies production of hadronic states @xmath37 that do not exclusively carry photon quantum numbers .
it is , accordingly , not surprising that the elastic diffraction obtained from ( [ ( 11 ) ] ) with the parameters employed for @xmath33 underestimates the measured cross section considerably , in particular for high values of the mass @xmath36 of the state @xmath37 .
compare ref.@xcite for a comparison with the zeus data @xcite .
theoretical approaches @xcite to the description of high - mass diffractive production frequently introduce a quark - antiquark - gluon @xmath43 component in the incoming photon . as this component
is usually ignored @xcite in the treatment of the total cross section , i am afraid , there is the danger of an inconsistency , due to a violation of the optical theorem
. a consistent inclusion of the @xmath44 component in elastic diffraction is contained in ref.@xcite , while an attempt for a consistent and unified treatment of inelastic and elastic diffraction and the total cross section , is provided in ref.@xcite .
i return to the analysis of the total cross section .
the above discussion of diffraction , in particular the sum rule ( [ ( 13 ) ] ) , suggests to introduce an upper limit @xcite in the integration over @xmath45 in @xmath46 . at finite energy @xmath32 ,
the diffractively produced mass spectrum is undoubtedly bounded by an upper limit that increases with energy . in our previous analysis @xcite
, we ignored such an upper limit , since the contribution of high masses seemed to be suppressed anyway .
we have examined the effect of a cut - off , @xmath47 , in the momentum space version of ( [ ( 4 ) ] ) or , equivalently , in ( [ ( 13 ) ] ) .
putting m^2_1 = ( 22 gev)^2 = 484 gev^2 , [ ( 15 ) ] that is the mass of the largest bin in the zeus data @xcite , we obtain an excellent description of all data with @xmath4 , all @xmath5 , as shown in fig . 1 .
putting @xmath48 overestimates the cross section @xmath33 significantly for @xmath49 , while values of @xmath47 smaller than the upper bound ( [ ( 15 ) ] ) yield results below the experimental ones at large @xmath13 .
it is gratifying that the simple procedure of introducing a cut - off for @xmath50 ( compare fig .
1 ) that may presumably be avoided by a refined treatment . ]
that ( aproximately ) coincides with the upper limit for diffractive production extends the gvd / cdp to the full region , as assumed in our previous analysis @xcite , the restriction of @xmath51 had to be imposed . ] of @xmath4 , all @xmath5 , where diffraction dominates the virtual compton - forward - scattering amplitude .
compare ref.@xcite for a comparison of the gvd / cdp with data for the longitudinal cross section from an h1 analysis . in conclusion :
+ i ) scaling , @xmath52 , in @xmath13 yields @xmath53 for @xmath2 at fixed @xmath3 and provides evidence for saturation .
\ii ) sum rules relate the elastic component in diffractive production to the total cross section , the terminology gvd / cdp being appropriate for low - x dis .
\iii ) the excess of diffractive production over the elastic @xmath54 component is presumably due to higher spin components , and accordingly \iv ) any theory of diffraction has to discriminate between an inelastic and an elastic component and must be examined with respect to its compatability with the total cross section , @xmath33 .
|
the evolution and maintenance of sexual replication is one of the central questions in modern evolutionary biology @xcite .
the theories with the broadest acceptance are that sex allows for the removal of deleterious mutations from a population @xcite , or that sex allows for faster adaptation in dynamic environments @xcite .
as has been noted in previous studies @xcite , the various theories for the existence of sex are incomplete , in that they do not explain why some organisms are obligately sexual , while other organisms either alternate between asexual and sexual replication , or are asexual replicators with some ability for recombination with other organisms . in a recent set of papers @xcite , tannenbaum , fontanari , and lee studied the competition between asexual and sexual replicators in various regimes .
two of the models consider single - celled organisms , that replicate by dividing into haploids .
these haploids then enter a haploid pool , where they fuse with other haploids .
the resulting diploid then divides through the normal mitotic process .
two other models were developed to be more appropriate for modeling multicellular organisms , which release either asexual spores or gametes . here , we considered the case where the population produces identical gametes , and where the population produces distinct sperm and egg gametes . for both sets of models it was found that sexual replication is favored in slowly replicating organisms and high population densities . for the case of multicellular organisms , it was found that distinct sperm and egg gametes were necessary to maintain a selective advantage for sexual replication over asexual replication @xcite . in all the models considered
, the authors assumed that there is a time cost for sex , due to the time it takes for a given haploid to find another haploid with which to fuse .
when the time cost for sex is low , then the selective advantage for sex outweighs the fitness penalty .
once the time cost for sex becomes sufficiently large , the selective advantage for sex no longer outweighs the fitness penalty , and asexual replication becomes the preferred replication strategy . in order to facilitate an exact analysis of the models , the authors made a number of simplifiying assumptions
: it was assumed that the organisms have diploid genomes consisting of only two chromosomes , and that each chromosome is only considered functional , or viable , if it is equal to some master sequence .
it was also assumed that the fitness of the organisms is determined by the number of viable chromosomes in the genome ( @xmath1 , @xmath2 , @xmath3 ) , and that organisms with two defective chromosomes have zero fitness .
finally , one of the key simplifications that was made in the sexual replication models was the assumption that only haploids with viable chromosomes can fuse with one another .
the reasoning behind this selective mating strategy was that haploids with defective chromosomes are not viable , and simply can not participate further in the replication process .
while the assumption of a selective mating strategy is the easiest one to study , and while it is broadly consistent with the observation that organisms do engage in mate selection , it is nevertheless an overly restrictive assumption that must be relaxed if one wants to develop more realistic sexual replication models .
the reason for this is that it is impossible , in practice , for one organism to read another organism s genome .
even if this were possible , the correlation between the genome sequence and fitness is extremely difficult to obtain .
organisms that engage in mate selection look for certain behaviors and physical attributes , known as _ indicator traits _ , that suggest that the given organism has a good genome @xcite . however , the map from such traits to genome is not exact .
furthermore , the more accuracy one wants in assessing the fitness of an organism , the more time one has to spend studying the organism . for
, if assessing the fitness of an organism is equivalent to reading its genome , then the more accuracy one wants , the more of the genome that has to be read .
therefore , a selective mating strategy has an additional time cost over other mating strategies , leading to a fitness penalty that may eliminate the advantage for sexual replication entirely .
this time cost was not explicitly considered in previous studies .
even if the time cost for a selective mating strategy were explicitly considered , we have shown that there will always be an uncertainly as to the exact fitness of the organism . as a result ,
more realistic models for sexual replication need to consider mating strategies that take into account the uncertainty that one organism has of another s fitness .
otherwise , the conclusions that we have drawn regarding the selective advantage for sex are open to the criticism that they are based on an unrealistic and overly restrictive mating strategy . in this paper ,
we take the opposite extreme from a selective mating strategy , and consider sexual replication with a random mating strategy .
that is , we assume that all haploids participate in the replication process , and fuse with one another at random .
as long as the cost for sex is negligible , we find that , even with this non - selective mating strategy , the mean fitness for a sexually replicating population is greater than that of the corresponding asexually replicating population . nevertheless , if the time cost associated with the selective mating strategy is negligible as well , then the mean fitness of the selective mating strategy is greater than that of the random mating strategy , independent of the cost for sex . as the cost for sex increases , sexual replication via either the selective or random mating strategies only outcompetes asexual replication over progressively smaller ranges of replication fidelities , and ceases to be advantageous entirely once the cost for sex crosses a threshold value .
this paper is organized as follows : in the following section ( section ii ) , we develop and analyze the mutation - selection equations appropriate for analyzing an asexually replicating , two - chromosomed , diploid population . in section iii
, we develop the mutation - selection equations appropriate for analyzing the sexually replicating analogue of the population considered in section ii .
we consider both selective and random mating strategies .
although the selective mating strategy has already been studied , we review it here for the sake of completeness and consistency of notation . in section iv , we compare all three replication strategies , and show that a random mating strategy outcompetes asexual replication when the cost for sex is negligible .
we also show that when the cost of implementing the selective mating strategy is negligible , then the selective mating strategy outcompetes the random mating strategy , independently of the cost for sex .
section v is the discussion , where we review and discuss the implications of our results . finally , in section vi we conclude the paper with a summary of the main results of the paper , as well as our plans for future research .
we consider a population of unicellular organisms whose genomes consist of two chromosomes .
we assume that a given chromosome , denoted @xmath4 , is functional if and only if it is equal to some master " , or wild - type , chromosome @xmath0 .
assuming first - order exponential growth , we then assume that the fitness , or first - order growth rate constant , of a given genome is determined by whether the genome consists of zero , one , or two functional chromosomes . to this end , we let @xmath5 denote the first - order growth rate constant of organisms with two functional , or equivalently , viable chromosomes . we let @xmath6 denote the first - order growth rate constant of organisms with one functional and one non - functional ( unviable ) chromosome . finally , we let @xmath7 denote the first - order growth rate constant of organisms with two non - functional chromosomes .
we assume that @xmath8 .
we will also assume that @xmath9 , which makes sense , since an organism with two defective chromosomes is not expected to grow .
we also define @xmath10 , and @xmath11 .
we now divide the populations into three subpopulations : we let @xmath12 , @xmath13 , and @xmath14 denote the number of organisms with two , one , and zero viable chromosomes .
we let @xmath15 be the total number of organisms , so that @xmath16 .
the population fractions @xmath17 , @xmath18 , and @xmath19
are then defined via @xmath20 , @xmath21 , and @xmath22 . to develop the mutation - selection equations describing the evolutionary dynamics of the asexually replicating equations
, we assume that the replication of each cell occurs as follows : the two chromosomes line up along the center of the cell and replicate ( see figure 1 ) . because replication is in general not error - free ,
we let @xmath23 denote the probability that a given genome is replicated correctly .
if the genome is sufficiently long , then the probability that a mutation will occur in a previously mutated region of the genome is negligible , so that an unviable chromosome produces an unviable daughter with probability @xmath2 .
this assumption is known as the _ neglect of backmutations_. we also let @xmath24 denote a co - segregation parameter , which is simply the probability that a parent chromosome co - segregates with the other parent chromosome in the cell .
figure 1 illustrates the various parent cell configurations and the final daughter cell configurations , along with their associated probabilities .
it should be noted that @xmath25 is the same as the @xmath26 parameter defined in @xcite . with these definitions in hand ,
we may develop expressions for @xmath27 , @xmath28 , and @xmath29 .
changing variables from population numbers to population fractions , we obtain the mutation - selection equations , @xmath30 x_{vv } \nonumber \\ & & \frac{d x_{vu}}{dt } = [ \kappa_{vu } p - \bar{\kappa}(t ) ] x_{vu } + 2 b_{vv } \kappa_{vv } x_{vv } \nonumber \\ & & \frac{d x_{uu}}{dt } = [ \kappa_{uu } - \bar{\kappa}(t ) ] x_{uu } + ( 1 - p ) \kappa_{vu } x_{vu } \nonumber \\ & & + s ( 1 - p)^2 \kappa_{vv } x_{vv } \nonumber \\\end{aligned}\ ] ] where @xmath31 , and , @xmath32 \nonumber \\ & & b_{vv } = [ 1 - p ] [ 1 - s ( 1 - p)]\end{aligned}\ ] ] the quantity @xmath33 is the mean fitness of the population , since it measures the first - order growth rate of the population as a whole . in order to determine which replication strategy is advantageous in a given regime , we compare the steady - state mean fitnesses of the populations employing the various strategies .
the population with the largest mean fitness will drive the others to extinction , and so the corresponding replication strategy is the advantageous one .
from quasispecies theory , it may be shown that the above system of equations converges to a steady - state , and that the steady - state mean fitness is given by @xmath34 , assuming that @xmath9 @xcite .
this implies that there exists a @xmath35 $ ] such that @xmath36 for @xmath37 $ ] , and @xmath38 for @xmath39 $ ] .
if we define @xmath40 , then we have that @xmath41 for @xmath37 $ ] , and @xmath42 for @xmath39 $ ] , where @xmath43 is defined by the equality @xmath44 .
note that @xmath45 is simply the steady - state mean fitness of the asexual population , normalized by the fitness of the wild - type , @xmath5 .
when we analyze the sexual replication models , we will also work with the normalized mean fitnesses , as it will prove convenient to do so . when comparing the various replication strategies , the strategy with the largest normalized mean fitness at steady - state is the one that will outcompete the others for the given set of parameters .
the general sexual replication model we are considering is illustrated in figure 2 : a diploid cell grows to mature size , with a first - order growth rate constant given by @xmath5 , @xmath6 , or @xmath9 , depending on whether the genome has two , one , or zero functional chromosomes , respectively .
the mature diploid then divides into two haploids , which enter a haploid pool . the haploids fuse with one another , and the resulting diploids
then immediately divide via the normal mitotic process .
it should be noted that the corresponding first - order growth rate constants for the asexual and sexual populations are taken to be equal , since the first - order growth rate constants measure the characteristic time it takes a diploid to double in size before dividing . with sexual replication
, it is necessary to keep track of the haploid as well as the diploid population .
therefore , in addition to the quantities @xmath12 , @xmath13 , and @xmath14 defined in the previous section , we also have the quantities @xmath46 and @xmath47 , corresponding to the number of viable and unviable haploids , respectively .
the haploid fusion process is modeled as a binary collision reaction characterized by second - order rate constants @xmath48 , @xmath49 , and @xmath50 , corresponding to the @xmath51 , @xmath52 , and @xmath53 haploid collisions .
it should be noted that @xmath48 and @xmath50 are defined so that @xmath54 and @xmath55 are the rates of disappearance of the @xmath56 haploids and @xmath57 haploids respectively , due to @xmath51 and @xmath53 haploid fusion respectively .
the quantity @xmath49 is defined so that @xmath58 is the rate of disappearance of either the @xmath56 haploids or the @xmath57 haploids due to @xmath52 haploid fusion .
we also assume that the system volume increases so as to maintain a constant density of genomes in the population .
that is , if @xmath59 denotes the volume of the system , then we assume that @xmath60/v $ ] is constant .
we now make the following definitions : we define @xmath16 to be the total population of diploids , and @xmath61 to be the mean fitness of the diploid population .
we define the population ratios @xmath20 , @xmath21 , @xmath22 , @xmath62 and @xmath63 .
finally , we define a diploid density @xmath64 .
if we write down the differential equations governing the values of @xmath12 , @xmath13 , @xmath14 , @xmath46 and @xmath47 , then changing variables to the population ratios gives the mutation - selection equations , @xmath65 where @xmath66 and where we have defined the additional quantities @xmath67 , @xmath68 , and @xmath69 via , @xmath70
although we are defining the mean fitness of the population with respect to the diploid organisms , at steady - state any two mean fitnesses defined with respect to two distinct sets of subpopulations will be equal .
the reason for this is that , at steady - state , the population reaches a mutation - selection balance , so that any two given sets of subpopulations will be in a fixed proportion to one another ( e.g. the total diploid population versus the total population ) .
therefore , the per capita rate of increase of one subpopulation is equal to the per capita rate of increase of another . as a result , there is no ambiguity as to which sexual mean fitness to use when comparing whether a given sexual population outcompetes another sexual population , or an asexual population .
the selective mating strategy is defined by @xmath71 , @xmath72 .
this implies that only the @xmath56 haploids are allowed to mate , while the @xmath57 haploids are essentially thrown away .
one justification for this mating strategy is that the @xmath57 haploids contain defective genomes , and so are simply physically incapable of participating further in the replication process .
another justification , one that is likely more relevant to actual organisms , is that the @xmath56 haploids have a way of determining the fitness of a potential haploid mate ( via indicator " traits , for example ) , and choose to only mate with the fittest haploids . for the selective mating strategy ,
the steady - state equations are given by , @xmath73 where @xmath74 .
we purposely neglect the steady - state equation corresponding to @xmath75 , since this equation will not be necessary to determine the mean fitness of the population .
solving the last equation for @xmath76 and substituting the result into the expression for @xmath77 gives , @xmath78 and so , @xmath79 we also have , @xmath80 substituting these values into the expression for @xmath77 , and making use of the fact that @xmath81 gives , after some manipulation , @xmath82 where @xmath83 is the normalized steady - state mean fitness of the population , and is given by @xmath84 . here
, @xmath85 stands for selective sexual " .
the steady - state equations for @xmath17 and @xmath18 then give , @xmath86 therefore , @xmath87 , so that @xmath88 now , the first steady - state equation for @xmath17 may be solved to give , @xmath89 , which may be substituted into eq .
( 10 ) to give , @xmath90 substituting in the value for @xmath17 gives , after some manipulation , that , @xmath91 ^ 2}{\phi_{ss } ( \phi_{ss } + 1)^2 ( \phi_{ss } + \alpha)^2 } = \frac{1}{2 } \frac{\kappa_{vv}}{\gamma \rho}\ ] ] we should note that the results we have obtained here for the selective mating strategy were previously derived by tannenbaum and fontanari @xcite .
nevertheless , we reviewed them here for the sake of completeness .
the random mating strategy is defined by @xmath92 . here , any haploid pair is equally likely to fuse as any another haploid pair .
the steady - state equations are , @xmath93 where @xmath94 .
the third equation is obtained by adding the equations corresponding to @xmath95 .
we then have , @xmath96 so that , @xmath97 if we define @xmath98 , @xmath99 , then , @xmath100 and so , @xmath101 by substituting the expression for @xmath102 into the expression defining @xmath77 , we obtain , after some manipulation , @xmath103 where @xmath104 . here
, @xmath105 stands for random sexual " .
the steady - state equations for @xmath17 and @xmath18 give , @xmath106 now , from eq .
( 16 ) it can be seen that @xmath107 , and so from eq .
( 20 ) we have , @xmath108 where @xmath109 , @xmath110 .
we now have , @xmath111 the second equation gives ,
@xmath112 \nonumber \\\end{aligned}\ ] ] now , @xmath113 , and so , after some manipulation , we obtain , @xmath114/ \nonumber \\ & & [ \phi_{rs } f(\phi_{rs } , \lambda ) ( 1 - 2 \alpha ) \nonumber \\ & & + \alpha ( \frac{1}{2 } b_{vu } ( 1 + f(\phi_{rs } , \lambda ) ) - f(\phi_{rs } , \lambda ) ) ] \nonumber \\\end{aligned}\ ] ] where we made use of the fact that @xmath115 and @xmath116 . plugging the value of @xmath17 back into the first equation from eq .
( 23 ) we obtain , after tedious algebra , @xmath117 \times \nonumber \\ & & [ \phi_{rs } f(\phi_{rs } , \lambda ) - \alpha(\frac{1}{2 } b_{vu } ( 1 + f(\phi_{rs } , \lambda ) ) - f(\phi_{rs } , \lambda ) ) ] \times \nonumber \\ & & [ \phi_{rs } f(\phi_{rs } , \lambda ) ( 1 - 2 \alpha ) + \alpha ( \frac{1}{2 } b_{vu } ( 1 + f(\phi_{rs } , \lambda ) ) - f(\phi_{rs } , \lambda ) ) ] \nonumber \\ & & = a_{vv } ( 1 - \alpha)^2 f(\phi_{rs } , \lambda ) ( 1 + f(\phi_{rs } , \lambda ) ) \phi_{rs}^2\end{aligned}\ ] ] now , @xmath118 , and so , multiplying both sides by @xmath119 gives , @xmath120 where , @xmath121 the solution that is chosen is the one that gives @xmath122 as @xmath123 , since @xmath124 is the equation defining the steady - state mean fitness when there is no cost for sex .
we now compare the various replication strategies .
we consider first the case where there is no cost for sex , so that @xmath125 , followed by the case where there is a non - zero cost for sex , so that @xmath126 . the ratio @xmath127 measures the cost for sex because it may be interpreted as the ratio of the characteristic time a haploid spends looking for another haploid with which to fuse , which is on the order of @xmath128 , to the characteristic time it takes newly formed diploid cell to grow to maturity and divide , which is on the order of @xmath129 .
when this ratio is small , then the fraction of the organism s life cycle that is devoted to the haploid fusion process is small , so that the time cost associated with sex is small .
conversely , when this ratio is large , then the time cost associated with sex is large as well .
when @xmath125 , the normalized mean fitness for the population replicating with the selective sexual replication strategy is given by , @xmath130 while the normalized mean fitness for the population replicating with the random sexual replication strategy is given by , @xmath131 the central result of this subsection is that @xmath132 , except when @xmath133 , @xmath134 , @xmath135 , or @xmath136 , in which case @xmath137 . we will prove this result in two steps : first we will prove that @xmath137 for @xmath133 , @xmath134 , @xmath135 , or @xmath136 .
we will then prove that @xmath132 as long as @xmath138 .
when @xmath133 , @xmath83 , @xmath141 are obtained by solving , @xmath142 which have the solutions @xmath143 .
we choose these solutions , because they are the ones that are physical . when @xmath134 , @xmath83 , @xmath141 are obtained by solving , @xmath144 so that @xmath145 .
when @xmath135 , we obtain , @xmath146 these two equations both admit the solutions @xmath1 and @xmath147 . since @xmath148 for @xmath134 , and @xmath149 for @xmath133 , by continuity it follows that @xmath150 for @xmath37 $ ] , and @xmath1 for @xmath39 $ ] .
when @xmath136 , we obtain , @xmath151 so that @xmath152 . note that in all cases , we have @xmath153 , as we wished to show . when @xmath138 , we claim that eq . ( 30 ) has a solution in @xmath154 . by continuity , we expect that this solution is the value of @xmath141 as a function of @xmath23 , since it is the solution that is consistent with @xmath155 at @xmath133 and @xmath156 at @xmath134 . when @xmath157 , eq .
( 30 ) evaluates to @xmath158 . when @xmath159 , eq .
( 30 ) evaluates to @xmath160 . by the intermediate value theorem
, it follows that eq .
( 30 ) has a solution in the interval @xmath154 .
this of course shows that @xmath161 .
we now claim that @xmath162 . for , if @xmath163 , then from eq .
( 30 ) we have , @xmath164 which implies that @xmath159 . but
this means that @xmath165 , since @xmath166 by assumption .
therefore , @xmath162 , as claimed .
if we can now show that there exists a @xmath166 such that @xmath167 , then we will have proven that @xmath168 for all @xmath166 . for otherwise , by the intermediate value theorem we would be able to find a @xmath166 such that @xmath169 .
now , @xmath43 is defined by the equation @xmath170 , so that @xmath171 and @xmath172 . since we are assuming @xmath173 , we either have @xmath174 or @xmath175 . if @xmath174 , then @xmath176 , so since @xmath177 at @xmath178 , we have that @xmath167 at @xmath179 , thereby proving that @xmath167 for @xmath166 . taking the limit @xmath180 gives that @xmath181 for @xmath166 , which of course implies that @xmath167 for @xmath166 , since @xmath162 for @xmath166 .
we have now shown that @xmath182 when @xmath138 , and so @xmath183 for @xmath138 .
we now turn to proving that @xmath184 for @xmath138 .
if we define @xmath185 , then @xmath186 .
if @xmath187 for some @xmath166 , then defining @xmath188 , we obtain from eq .
( 30 ) that , @xmath189 \nonumber \\ & & \rightarrow [ 2 a_{vv } ( 1 - \alpha)^2 \omega - ( \omega + 1 - \alpha p)((1 - 2 \alpha ) \omega + \alpha p ) ] \phi \nonumber \\ & & = ( \omega + 1 - \alpha p ) ( 1 - 2 \alpha ) \alpha p - 2 a_{vv } ( 1 - \alpha)^2 \alpha p\end{aligned}\ ] ] now , @xmath190 , so that , @xmath191 \phi = -\alpha p\ ] ] so , multiplying eq . ( 29 ) by @xmath192 gives , @xmath193 which is impossible for @xmath138 .
therefore , @xmath194 for @xmath166 .
now , let us look at @xmath195 and @xmath196 at @xmath134
. for @xmath83 we have , @xmath197 while for @xmath141 we have , @xmath198 therefore , for @xmath23 near @xmath2 , we have , @xmath199 so since @xmath200 , it follows that @xmath184 for @xmath201 , but @xmath23 close to @xmath2 . but then @xmath184 for all @xmath166 , for otherwise there exists a @xmath166 for which @xmath202 , which implies by the intermediate value theorem that there exists a @xmath166 such that @xmath203 .
we have now established that @xmath204 for @xmath138 , as we set out to prove . the central result of this subsection is that @xmath184 for @xmath138 , with equality only occurring when @xmath133 , @xmath134 , @xmath135 , or @xmath136 .
suppose we can show that @xmath194 for @xmath138 , independently of the value of @xmath127 .
then if @xmath202 for some @xmath138 and for some value of @xmath205 , it follows from continuity and the intermediate value theorem that @xmath206 for some value of @xmath205 , since @xmath207 for @xmath138 when @xmath125 .
this is a contradiction , and so we must have that @xmath184 for all @xmath138 and @xmath205 . therefore , to prove the central result of this subsection , we will show that @xmath206 for @xmath139 and/or @xmath140 , and that @xmath194 for @xmath138 .
working again with the definition @xmath208 , we have from eq . ( 14 ) that , @xmath209 as @xmath210 increases from @xmath1 to @xmath211 , we expect @xmath83 to decrease from its maximal value down to @xmath1 . in this regime
, we would like to determine the sign of the expression @xmath212 .
consider the polynomial @xmath213 , which has the roots @xmath214 $ ] .
if @xmath215 , then the + " root , denoted @xmath216 , is positive , and the - " root , denoted @xmath217 , is negative .
therefore , writing @xmath218 , we may note that when @xmath219 $ ] , @xmath220 and @xmath221 , so that @xmath222 . the positive root of @xmath223 is the value of @xmath83 when @xmath224 .
since @xmath83 then decreases to @xmath1 as @xmath210 increases , it follows from our analysis that @xmath223 is zero or negative for @xmath225 .
therefore , taking the square roots of both sides of eq .
( 42 ) gives , @xmath226 if @xmath206 for a given @xmath227 $ ] and @xmath225 , then we may insert the expression above into eq .
( 27 ) defining @xmath141 .
if we set @xmath228 , then we obtain , after some manipulation , @xmath229 \times \nonumber \\ & & [ \phi^2 - ( 2 a_{vv } ( 1 - \alpha ) - 1 + \alpha p ) \phi - \alpha p]^2 \nonumber \\ & & - 2 [ \phi + 1 ] \times \nonumber \\ & & [ ( 1 - 2 \alpha ) \phi^2 - \alpha^2 ( 1 - p ) \phi + \alpha^2 p ] \times \nonumber \\ & & [ \phi^2 - ( 2 a_{vv } ( 1 - \alpha ) - 1 + \alpha p ) \phi - \alpha p ] \nonumber \\ & & + 2 a_{vv } ( 1 - \alpha)^2 \phi^2 [ \phi^2 - ( 2 a_{vv } ( 1 - \alpha ) - 1 + \alpha p ) \phi - \alpha p ] \nonumber \\ & & - 2 a_{vv } ( 1 - \alpha)^2 \phi^2 [ \phi + 1 ] [ \phi + \alpha ] \nonumber \\ & & + [ \phi + 1]^2 [ \phi + \alpha ] [ \phi - \alpha p][\phi ( 1 - 2 \alpha ) + \alpha p ] = 0\end{aligned}\ ] ] this expression may be simplified ( with the aid of a symbolic math package if necessary ) to give , @xmath230 \nonumber \\\end{aligned}\ ] ] which implies that either @xmath140 , @xmath134 , or @xmath231 . since @xmath232 is equivalent to @xmath133 when @xmath233 , and since @xmath234 is equivalent to @xmath134 and @xmath224 , we obtain that @xmath206 only when @xmath140 or @xmath139 .
if @xmath138 , then @xmath194 . therefore , we have proven that @xmath206 when @xmath133 , @xmath134 , @xmath135 , or @xmath136 , and @xmath184 for @xmath138 , independently of the value of @xmath127 .
the two key results of this paper are that a sexual population employing a random mating strategy will outcompete an asexual population when the cost for sex is negligible , and that a sexual population using a selective mating strategy will outcompete a sexual population using a random mating strategy .
the only exceptions are the boundary cases @xmath139 and @xmath140 .
however , even here , the mean fitnesses of the two sexual strategies are identical .
furthermore , when the cost for sex is negligible , then the mean fitnesses of the sexual strategies are identical to that of the asexual strategy .
that random mating provides a selective advantage over asexual replication is an interesting result , because the strategy can lead to the formation of diploids with completely defective genomes .
presumably , however , the fitness benefit provided by the formation of diploids with two functional chromosomes outweighs the fitness cost associated with the formation of diploids with two defective chromosomes , leading to an overall advantage for the strategy . nevertheless ,
because the selective mating strategy does not produce genomes with defective chromosomes , this strategy has an advantage over the random mating strategy . when the cost for sex is negligible , an analysis of the mean fitnesses @xmath45 , @xmath83 , and @xmath141 near @xmath134 yields some interesting results .
we have @xmath235 .
we have also shown that when @xmath23 is close to @xmath2 , then to first - order in @xmath236 we have @xmath237 and @xmath238 .
therefore , we may note that the random mating strategy and the asexual strategy are identical to first - order in @xmath236 , while the selective mating strategy already outcompetes both the random and asexual strategies . however , from eq .
( 30 ) we obtain , @xmath239 and so , to second - order in @xmath236 we have that , @xmath240 ( 1 - p)^2\ ] ] comparing the second - order expression for @xmath141 to @xmath45 , we see that @xmath141 exceeds @xmath45 by @xmath241 when @xmath23 is close to @xmath2 .
figure 3 shows a plot of the two sexual replication strategies when there is no cost for sex and the asexual strategy .
when there is a cost for sex , then when @xmath134 the asexual population outcompetes both sexual populations .
this makes sense , for when @xmath134 replication is error - free , and hence at steady - state the asexual population consists only of the wild - type . in this case
, genetic recombination will not improve fitness , since there are no defective chromosomes in the population to begin with .
however , because sexual replication when there is no cost for sex will outcompete an asexual replication when @xmath166 , it follows that if the cost for sex is sufficiently low , then below a certain value of @xmath23 a sexual population will outcompete the asexual population . presumably , the higher the cost for sex , the smaller @xmath23 must be before the selective advantage for sexual replication is sufficiently large to outweigh the cost .
this behavior only persists up to a maximal cost for sex , beyond which asexual replication outcompetes sexual replication at all replication fidelities .
the reason for this is that the sexual and asexual mean fitnesses converge to @xmath1 as @xmath242 . as a result ,
once the cost for sex is sufficiently high , the fitness advantage of the sexual strategy for the values of @xmath23 where the sexual strategy can outcompete the asexual strategy is too small to overcome the cost for sex . indeed , from eq .
( 14 ) it may be shown that @xmath243 when @xmath126
. since @xmath244 , it follows for @xmath173 that @xmath245 in a neighborhood of @xmath23 sufficiently close to @xmath1 .
since @xmath246 , the same condition holds for @xmath141 as well .
however , if any of the sexual populations can outcompete the asexual population when @xmath23 is sufficiently large , then there must exist another value of @xmath23 below which asexual replication outcompetes sexual replication , and above which sexual replication outcompetes asexual replication .
the complete picture is then one where , for a non - zero cost for sex , the asexual population outcompetes a sexual population above a certain replication fidelity .
below this replication fidelity , the sexual population outcompetes the asexual population .
finally , once the replication fidelity becomes sufficiently low , the asexual population again outcompetes the sexual population . based on this analysis
, we expect that , as the cost for sex increases from @xmath1 to @xmath211 , the region of replication fidelities where sexual replication outcompetes asexual replication starts at @xmath247 , gradually shrinks , and eventually disappears once the cost for sex crosses a threshold value . of course , because the random mating strategy has a lower fitness than the selective mating strategy
, the region of replication fidelities where the random mating strategy outcompetes asexual replication is a subset of the region of replication fidelities where the selective mating strategy outcompetes asexual replication .
figure 4 shows a plot of the asexual and sexual mean fitnesses when there is a non - zero cost for sex .
figure 5 shows a plot of the regions for the selective advantages of the various replication strategies as a function of @xmath127 .
although we have shown that a selective mating strategy will outcompete a random strategy , this analysis is based on the assumption that a selective mating strategy is not inherently slower than a random mating strategy .
as discussed in the introduction , this is in general not true , since there is a time cost associated with determining the genome of a potential haploid mate .
this time cost leads to an additional fitness cost associated with a selective mating strategy that is not incurred by a random mating strategy .
when the replication fidelity is either sufficiently high or sufficiently low , then the benefits of a selective mating strategy are not sufficient to overcome the fitness costs , and so the random mating strategy dominates . however , at intermediate replication fidelities , the selective mating strategy may indeed outcompete the random mating strategy , assuming that the fitness cost associated with selective mating is not too high .
once the fitness cost associated with selective mating becomes sufficiently high , then the random mating strategy may outcompete the selective mating strategy at all replication fidelities . in summary , when the time cost for a selective mating strategy is taken into account
, an analysis of the regimes where the selective and random mating strategies are expected to be respectively dominant may produce a curve that is analogous to the ones in figure 5 .
this paper developed a set of simplified models describing asexual and sexual replication in a unicellular population consisting of two - chromosomed , diploid genomes .
we considered two types of sexual replication strategies : a selective mating strategy , where only viable haploids are allowed to fuse , and a random mating strategy , where all haploids are allowed to participate in the replication process .
we assumed that haploid fusion was a second - order rate process .
we found that , when the cost for sex is negligible , both the selective and random mating strategies lead to a greater mean fitness than asexual replication .
nevertheless , we found that the selective mating strategy has a higher mean fitness than the random mating strategy , as long as the additional fitness penalty associated with a selective mating strategy is negligible .
further analysis suggested that sexual replication for both the selective and random mating strategies is favored at intermediate mutation rates , and when the cost for sex is sufficiently low .
once the cost for sex becomes sufficiently high , the selective advantage for sexual replication disappears entirely .
the results of this paper therefore suggest that sex is favored in slowly replicating organisms and high population densities . while this is consistent with previous studies @xcite , what is interesting is that this result holds even with a random mating strategy . for future work , we will consider models where it is not necessarily true that @xmath9 , but rather we will allow for positive values of @xmath7 .
this will make our model more consistent with quasispecies models that often assume a small , but positive growth rate for organisms with defective genomes .
furthermore , although our model provides results that are broadly consistent with actual organismal behavior , we have nevertheless worked with a highly simplified sexual replication model that does not exactly correspond to the sexual replication pathway employed by unicellular organisms . in our model
, a mature diploid divides into two haploids , the haploids fuse , and then the resulting diploid divides into two cells .
thus , the diploid mitosis occurs _
after _ haploid fusion .
by contrast , in _ saccharomyces cerevisiae _ ( baker s yeast ) , for example , a mature diploid divides into two diploids , which then divide into four haploids .
the haploids then fuse , and the resulting diploids grow to maturity to begin the process again .
thus , for yeast , diploid mitosis occurs _ before _
haploid fusion .
we suspect that the relative positions of the diploid mitosis and haploid fusion stages in a sexual replication cycle can affect the mutation - selection balance , and so we would like to develop more realistic models corresponding to the sexual replication pathways in actual unicellular organisms .
we should emphasize , however , that we believe that the central results of this paper will hold even when we consider a somewhat different sexual replication pathway .
finally , one of the simplifying assumptions made in this paper is that the organisms have genomes consisting of two chromosomes . for future research
, we plan to study genomes consisting of arbitrary numbers of chromosomes . within this context
, we plan to consider more complex fitness landscapes and the role of intra - genomic recombination .
this research was supported by a start - up grant from the united states
israel binational science foundation , and by an alon fellowship from the israel science foundation .
the author would also like to thank j.f .
fontanari for helpful conversations leading to the completion of this work .
e. tannenbaum , a comparison of three replication strategies in complex multicellular organisms : asexual replication , sexual replication with identical gametes , and sexual replication with distinct sperm and egg gametes , " submitted .
|
this paper develops a simplified set of models describing asexual and sexual replication in unicellular diploid organisms .
the models assume organisms whose genomes consist of two chromosomes , where each chromosome is assumed to be functional if it is equal to some master sequence @xmath0 , and non - functional otherwise .
the first - order growth rate constant , or fitness , of an organism , is determined by whether it has zero , one , or two functional chromosomes in its genome .
we assume that an organism with no functional chromosomes has zero fitness . for a population replicating asexually
, a given cell replicates both of its chromosomes , and then the cell divides and splits its genetic material evenly between the two cells . for a population replicating sexually , a given cell first divides into two haploids , which enter a haploid pool . within the haploid pool , haploids fuse into diploids , which then divide via the normal mitotic process .
haploid fusion is modeled as a second - order rate process .
we review the previously studied case of selective mating , where it is assumed that only haploids with functional chromosomes can fuse , and also consider the case of random haploid fusion .
when the cost for sex is small , as measured by the ratio of the characteristic haploid fusion time to the characteristic growth time , we find that sexual replication with random haploid fusion leads to a greater mean fitness for the population than a purely asexual strategy
. however , independently of the cost for sex , we find that sexual replication with a selective mating strategy leads to a higher mean fitness than the random mating strategy .
this result is based on the assumption that a selective mating strategy does not have any additional time or energy costs over the random mating strategy , an assumption that is discussed in the paper .
the results of this paper are consistent with previous studies suggesting that sex is favored at intermediate mutation rates , for slowly replicating organisms , and at high population densities .
|
the nonlinear effects are important in medical ultrasound and also plays a role in diagnostic program .
some of diagnostic ultrasound instruments have implemented harmonic imaging into their devices by receiving the harmonics in the reflected ultrasound caused by nonlinear distortion of the signal propagating through the biological tissue .
meanwhile , we should pay attention to observing hygienic limits of ultrasound exposition to avoid heating the unwanted tissue .
the ability to predict the effects of nonlinear ultrasound propagation therefore becomes important .
numerical simulations are currently the best means of making predictions of nonlinear ultrasound propagation .
two of the numerical techniques suitable for providing solutions to the propagation problem are the finite difference time domain(fdtd ) method@xcite and the discrete exterior calculus(dec)@xcite .
the westervelt equation is a model for the propagation of finite amplitude ultrasound , deriving from the equation of fluid motion by keeping up to quadratic order terms@xcite . in this paper , an conditional stable scheme for this equation using the techniques in dec
is proposed .
a significant advantage of this scheme is that it can be used to find numerical solutions on the discrete space manifold and the time , therefore is a generation of fdtd . in order to refine the space mesh to improve the accuracy of the solution given by the this scheme
, the amount of work involved increases rapidly , since the length of time step should also be reduced .
hence , two unconditional stable schemes are also proposed , namely the implicit and semi - implicit dec schemes for westervelt equation on manifold .
the @xmath0d or @xmath1d space manifold can be approximated by triangles or tetrahedrons , and the time by line segments .
suppose each simplex contains its circumcenter .
the circumcentric dual cell @xmath2 of simplex @xmath3 is @xmath4 where @xmath5 is all the simplices which contains @xmath3 , ... ,
@xmath6 , and @xmath7 is the circumcenter of @xmath5 .
a discrete differential @xmath8-form , @xmath9 , is the evaluation of the differential @xmath8-form on all @xmath8-simplices .
dual forms , i.e. , forms evaluated on the dual cell . in dec , the exterior derivative @xmath10 is approximated as the transpose of the incidence matrix of @xmath8-cells on @xmath11-cells , the approximated hodge star @xmath12 scales the cells by the volumes of the corresponding dual and primal cells , and the laplace operator is approximated as @xmath13 take fig.1 as an example for a part of 2d mesh , in which @xmath14 , ... , @xmath15 are vertices , @xmath16 , @xmath0 , @xmath1 are the circumcenters of triangles , @xmath17 , @xmath18 , @xmath19 are the circumcenters of edges .
denote @xmath20 as the length of line segment @xmath21 and @xmath22 as the area of quadrangle @xmath23 .
@xmath24{west / fvmdec1.eps } \end{center } \centering { figure.1 .
a part of 2d mesh } \end{minipage}\ ] ] define @xmath25 and @xmath26 in fig.1 , the discrete laplace operator acting on @xmath27 at vertex @xmath14 is @xmath28 @xmath24{west / juxing.eps } \end{center } \centering { figure.2 .
a part of 2d rectangular mesh } \end{minipage}\ ] ] in fig.2 , scheme ( 1 ) reduces to @xmath29where @xmath30 is the uniform space step length .
the weservelt equation can be written in the following form : @xmath31 where @xmath27 is the acoustic pressure , @xmath32 is the are the ambient density , and @xmath33 is the ultrasound speed , @xmath34 is the diffusivity of ultrasound , @xmath35 is the coefficient of nonlinearity . for some situations ,
a source having azimuthal symmetry about its axis is considered . in this case ,
2d triangular discrete manifold as the space is only need to be considered . and the eq.(2 ) on 3d space manifold and the time can also be solved numerically , using a similar approach . the temporal partial derivatives with discrete differences , which can be obtained from taylor series expansions about each node of the computational mesh .
temporal derivatives present in eq.(2 ) can be calculated as follows : @xmath36 where @xmath37 is the time step length , @xmath38 is the time coordinate .
let @xmath39 the explicit dec scheme for eq.(2 ) is @xmath40 if refining the space mesh , the amount of work involved increases rapidly , since the length of time step should satisfy the stable condition .
now , two unconditional stable schemes are proposed , namely the implicit dec scheme ( 4 ) and semi - implicit dec scheme ( 5 ) .
@xmath41 @xmath42 the higher order accuracy scheme for temporal derivatives can also be used in schemes(3 - 5 ) .
the first two terms in ( 2 ) describe linear wave propagation at the small - signal sound speed .
the third term describes the loss due to he viscosity and thermal conduction of the media .
the fourth term describes the nonlinear distortion of the traveling wave due to amplitude effects . since the coefficients of third and fourth terms in eq.(2 )
are small compared with the coefficient of second term , the effect of @xmath43 and @xmath44 on the stability of the scheme ( 3 ) can be ignored .
therefore , the stable condition for the wave equation @xmath45 is need to be satisfied .
the wave equation can be discretized as @xmath46 the stable condition condition for scheme ( 6)@xcite is @xmath47 in fig.2 , the inequality ( 7 ) reduces to @xmath48 which is the stable condition for scheme ( 6 ) on square grid . the analysis of unconditional stability for schemes ( 4 ) and ( 5 ) can also be proved as the wave equations . by the definition of truncation error ,
the exact solutions of westervelt equation satisfy the same relation as schemes ( 3 - 5 ) except for an additional term @xmath49 .
this expresses the consistency , and so the convergence for schemes ( 3 - 5 ) by lax equivalence theorem ( consistency @xmath50 stability @xmath51 convergence ) .
the derivative in eq.(2 ) is approximated by first order difference in schemes ( 3 - 5 ) .
equivalently , @xmath27 is approximated by linear interpolation function .
this is to say that schemes ( 3 - 5 ) have first order spacial and temporal accuracy , according to the definition about accuracy of finite volume method .
the dec schemes ( 3 - 5 ) for westervelt equation was implemented in c++ .
the implementation consists of the following steps : * set the simulation parameters .
these are the dimensions of the computational mesh and the size of the time step , etc . ;
* set the propagating media parameters .
* initialize the mesh indexes . *
assign current transmitted signal . *
compute the value of all spatial nodes and temporarily store the result in the circular buffer for further computation . *
visualize the currently computed grid of spatial nodes .
* repeat the whole process from the step 4 , until reach the desired total number of iterations . in the common practice ,
not every simulation step needs to be visualized , especially when the time step size is too small .
in all the following examples , scheme ( 3 ) is used , and the parameters are @xmath52 @xmath53 @xmath54 . the gaussian envelope @xmath55 is used as a source signal . the example in fig.3 shows the nonlinear propagation effects and the diffusivity in the propagation of the gaussian envelope .
@xmath24{west / tuzi1.eps } \includegraphics[scale=0.45]{west / tuzi2.eps } \includegraphics[scale=0.45]{west / tuzi3.eps } \end{center } \centering { figure.3 .
the propagation of gaussian envelope on a rabbit } \end{minipage}\ ] ] @xmath24{west / f1.eps } \includegraphics[scale=0.45]{west / f2.eps } \includegraphics[scale=0.45]{west / f3.eps } \end{center } \centering{figure.4 .
the propagation of gaussian envelope on a sphere with two kinds of media } \end{minipage}\ ] ]
|
the westervelt equation is a model for the propagation of finite amplitude ultrasound .
the method of discrete exterior calculus can be used to solve this equation numerically .
a significant advantage of this method is that it can be used to find numerical solutions in the discrete space manifold and the time , and therefore is a generation of finite difference time domain method .
this algorithm has been implemented in c++
. 0.2 cm * keywords : * westervelt equation , laplacian operator , discrete exterior calculus , manifold , numerical simulation . 0.2 cm * msc(2010 ) : * 35j05 , 65m12 , 65m08 , 53a25 .
|
we have initiated a systematic search program to find wolf - rayet ( wr ) stars of the carbon series ( wc subtypes ) in the so - called wr galaxies ( conti 1991 ) .
indeed , among the @xmath0 70 wr galaxies known today only few cases show the broad @xmath35808 emission originating from wc stars ( e.g. mrk 724 , ngc 4861 , ngc 4214 , he 2 - 10 , ngc 2363 ; cf .
references in meynet 1995 ) .
however , as meynet ( 1995 ) pointed out , a natural result of stellar evolution ( or our current knowledge of it ) is that one expects that in a large fraction of wr galaxies ( typically 30 % for metallicities @xmath4 ; cf .
schaerer & vacca 1996 ) the wr population should be dominated by stars of the wc subtype if star formation occurs on timescales short compared to the lifetimes of massive stars .
since the @xmath35808 feature is usually weaker than the characteristic wr bump at @xmath0 4700 it may simply have been overlooked in many previous observations .
interestingly , in the homogeneous sample of low metallicity galaxies of izotov , thuan & lipovetsky ( 1994 , 1997 ) , 5 out of 15 wr - rich objects show strong wc features , which corresponds surprisingly well to the theoretically predicted value of @xmath0 30 % .
apart from the implications for the understanding of massive star evolution , the presence or absence of a substantial population of wc stars in young starbursts may imply significant differences of the ionizing spectrum ( cf .
schaerer 1996 ) and the chemical enrichment due to massive stars ( cf .
maeder 1983 ) . as one of our targets we have observed the central region of the amorphous galaxy ngc 5253 at a distance of 4.1 mpc ( sandage et al .
this low metallicity object ( @xmath5 ) contains a very young starburst and thus provides an important laboratory for studies of local chemical enrichment , or `` chemical pollution '' in giant regions ( walsh & roy 1987 , 1989 , hereafter wr87
, wr89 ; pagel et al .
1992 ; esteban & peimbert 1995 ; kobulnicky et al .
1997 , hereafter ksrwr97 ) .
in addition , recent hst imaging ( meurer et al .
1995 ; gorjian 1996 ; calzetti et al .
1997 ) reveals numerous super star clusters in ngc 5253 and make this object an interesting place to study what might be proto - globular clusters ( cf .
conti & vacca 1994 ; meurer et al .
1995 , and references therein ) .
in this letter we report the detection of wr stars of both wn and wc subtype in two star forming regions of ngc 5253 .
the observations are presented in 2 . in 3 and
4 we determine the massive star content of the two regions and ages of the populations .
the implications regarding the source of ionization of ngc 5253 , the hardness of the ionizing flux , and the origin of n enrichment are discussed in 5 .
long - slit spectra of ngc 5253 were obtained on the night of 1995 april 24 25 at the eso 2.2 m telescope .
the data were acquired with the efosc2 spectrograph and a 1024@xmath61024 thomson ccd with a pixel size of 0.34 . during the photometric night
we also observed the spectrophotometric standard stars hd 84937 and kopff 27 in order to flux calibrate the spectra of the galaxy .
we used grism 4 which gives a spectral coverage of 4400 6500 with a resolution of @xmath0 5 .
the slit was oriented nne - ssw ( pa @xmath0 20@xmath7 ) in order to lie the brightest optical regions of the galaxy .
the slit width was 1.6 for the galaxy observations and 5 for the standard stars .
spectra of he - ar calibration lamp were obtained immediately before and after the galaxy integrations in order to accurately calibrate the wavelength scale .
the total integration time of 5400 s was divided into 5 exposures ( 4 times 20 minutes plus 10 minutes ) in order to avoid saturation of the bright nebular lines ( and [ ] ) and to recognize cosmic ray impacts .
the seeing was relatively stable during the observations with a mean spatial resolution of about 1 .
the spectrum was aquired at low airmass ( @xmath0 1.1 ) and no correction for the loss of blue light due to atmospheric dispersion was made .
the spectra were reduced according to standard reduction procedures using the midas package long .
these include bias subtraction , flat - field and airmass corrections , wavelength and flux calibrations , sky subtraction , and cosmic ray removal .
the spatial distribution of emission - lines along the slit reveals the presence of two bright regions , hereafter called _ a _ ( for the brightest ) and _ b _ ( south of _ a _ ) , separated by about 3 ( 60 pc at a distance of 4.1 mpc ) .
comparisons of our slit position with the hst images of meurer et al .
( 1995 , foc with f220w filter ) and gorjian ( 1996 , wfpc2 with f606w ) show that region _
a _ is located at the maximum of the optical emission ( = region 1 of gorjian ) , while _ b _ is centered on the maximum peak of uv emission ( = uv1 of ksrwr97 ) .
note that region 1 of wr89 ( = region a of wr87 ) contains both our regions _ a _ and _ b_. we also detected a region ( hereafter called _ h _ ) , located @xmath0 3 nne of _ a _ , for which no clear counterpart could be found in the hst images .
we extracted one - dimensional spectra corresponding to the different regions by adding 3 columns ( @xmath8 1 ) along the spatial dimension .
the two regions _ a _ and _ b _ are bright , low - metallicity regions with a large number of intense nebular emission lines seen in their spectra .
because of the medium spectral resolution , the emission lines in the wr bumps at @xmath0 4700 and 5800 ( see fig . [ fig_spectra ] ) and the [ ] / [ ] lines at 6300 are not well separated . in order to measure accurately the individual emission lines
, we use a procedure of multi - gaussians fitting where the number of gaussians is the only fixed parameter ( position , intensity and fwhm are free to vary ) .
line fluxes , fwhm and equivalent widths were determined using midas standard commands . due to the limited spectral range
, our observations do not include the h@xmath9 emission line commonly used to determine the internal reddening parameter @xmath10 . for regions _ a _ and _ h _ we therefore adopt @xmath11 0.44 and 0.85 respectively , taken at the appropriate locations from the extinction map of wr89 .
for _ b _ we adopt the reddening derived by ksrwr97 ( @xmath11 0.20 ) .
dereddened fluxes are derived using the extinction law of seaton ( 1979 ) including a galactic foreground extinction of e(b - v)=0.05 ( burstein & heiles 1984 )
. observed and dereddened fluxes are given in table 1 . the relative uncertainty of our measurements is smaller than 10% for the most intense lines ( and [ ] ) , and can reach about 40% for the faintest .
we note that our absolute flux calibration for region _ a _ agrees well with the values measured by campbell , terlevich & melnick ( 1986 ; hereafter ctm86 ) and wr89 .
up to three different broad emission lines ( fwhm @xmath12 30 ) have been detected with varying confidence levels in regions _ a _ and/or _ b _ over the wavelength range covered by our spectra : the / @xmath134640,4650 blend , @xmath34686 , and @xmath35808 - 5812 .
clear detections ( @xmath12 3 @xmath14 ) of broad lines are in _ a _ and _ b _ and @xmath34686 in _ b_. the detection and hence the fwhm of @xmath34686 in the spectrum of _ a _ dominated by high - excitation nebular emission lines of [ ] @xmath134711,4740 ( see fig .
[ fig_spectra ] ) is uncertain ( @xmath15 ) .
/@xmath134640,4650 may be present in _ a _ and _ b _ at a 1 - 2 @xmath14 level .
_ stands out with a very large @xmath34686/ratio , uncommon for the metallicity of ngc 5253 , and a relatively narrow @xmath34686 line ( fwhm=20 ) . due to these peculiarities and
the absence of a visible counterpart in any images ( cf . above ) this region is not discussed further .
future studies would be highly desirable .
the broad wr bump around 4700 has been previously detected by ctm86 , wr87 in their region 1 ( containing both our regions _ a _ and _ b _ ) , and in the region uv1
( = our @xmath16 region ) of ksrwr97 .
one of the main results of this letter is the unambiguous detection of a broad ( fwhm = 62 and 86 ) @xmath35808 emission line in regions _ a _ and _ b _ ( see fig .
[ fig_spectra ] ) , which clearly indicates the presence of wc stars in these regions ( see [ s_wrlines ] ) . a sufficiently high s / n is required for the detection of the @xmath3 5808 line in integrated spectra of extragalactic regions or alike objects , and to derive the observed frequency of wc stars in wr galaxies .
this explains the non detection of this line in previous ground - based spectra of ngc 5253 ( ctm86 , wr87 , wr89 ) and in the recent hst fos spectrum of the region uv1 of ksrwr97 .
interestingly , however , the presence of wc stars was suspected by wr89 based on a high carbon abundance derived from iue spectra ( cf .
both from the relative strength of @xmath34686 to @xmath35808 and from the large fwhm of the latter line , wn stars can be ruled out as the origin of the line .
the observed fwhm ( @xmath0 70 ) of this line corresponds well to early type wc stars ( typically wc4 ; smith , shara & moffat 1990 ) including possibly also some wo3 - 4 stars : their @xmath35590 emission ( @xmath17 15 % of @xmath35808 ) would remain undetectable . from the absence of @xmath35590 a significant number of wo1 - 2 stars can be excluded ( cf .
kingsburgh , barlow & storey 1995 ) .
similarly the absence of @xmath35696 excludes a significant population of late type wc stars .
the relative contributions of the / @xmath134640,4650 blend and @xmath34686 to the wr bump exclude a pure wc origin .
the fwhm ( @xmath0 30 ) of @xmath34686 agrees well with wnl stars ( smith , shara & moffat 1996 ) .
all broad emission line features therefore indicate a mixed population of wnl , and wc4 and/or wo3 - 4 stars in regions _ a _ and _ b_. this result is further supported by quantitative comparisons presented in 4 .
the presence of intermediate type wn / wc stars can not be excluded .
the approximate number of o and wr stars present in the observed regions can be estimated as follows . adopting the value @xmath18 from vacca & conti ( 1992 , hereafter vc92 ) for the lyman continuum luminosity of a o7v star and assuming case b recombination , we estimate @xmath0 865 ( 210 ) o7v stars in region _ a _ and in region _
b _ from their luminosity . these numbers are approximately @xmath0 5 and @xmath0 35 % lower , respectively , if we account for the contribution of the wr stars to the ionizing flux . depending on the age of the population and the imf slope
, the number of o7v stars does not necessarily correspond well to the total number of o stars : most likely the total number of o stars is larger by a factor of @xmath0 2 and 4 for _ a _ and _ b _ respectively ( schaerer 1996 ) . using the average observed luminosity of wc4 stars in the @xmath35808 line ( @xmath19 ; vc92 ) , the 5808 emission in regions a and b can be explained by 10 and 13 wc4 stars respectively . from the observed @xmath34686 line luminosity we derive an upper limit of 25 wnl stars for _ a _ ( depending on nebular contribution ) and 39 wnl stars for _ b _ , using the average line luminosity given by vc92 . to see whether the wnl , wc and o star populations are compatible with predictions from stellar evolution we use the recent evolutionary synthesis models of schaerer ( 1996 ) and schaerer & vacca ( 1996 ) , which allow direct comparisons to be made with the relevant observational quantities related to the wr and o star population ( fig . [ fig_models ] panels b - d ) .
the predictions are shown for instantaneous burst models at the appropriate metallicity for regions _ a _ and _ b _ ( z=0.004 , cf .
wr89 , ksrwr97 ) , assuming a salpeter imf with an upper mass cut - off of @xmath20 120 .
the observed line fluxes in the different wr lines ( in units of the flux ) are shown as filled symbols .
panel * a * illustrates the corresponding wr / o , wnl / o , and wc / o number ratios . figure [ fig_models ] shows that both the observed values of the wr lines fluxes and their variation with @xmath21 can be reasonably well reproduced by the models for z=0.004 . with several variations of the input parameters ( flatter imf , short burst duration , etc . )
all the observed values can be matched with greater accuracy . in view of the uncertainties of the measurement for these weak lines we conclude and all broad wr lines in both regions _ a _ and _ b _ are consistent with an approximately instantaneous burst .
the strong wr features also imply the presence of very massive stars ( @xmath22 60 ) . in the burst models a decrease of @xmath21 corresponds to an increasing age of the burst .
the ages derived from our models correspond to @xmath0 2.8 myr for _ a _ and 4.4 myr for _ b_. as shown above this age sequence is also compatible with the wr line fluxes , which exhibit changes on short time scales .
the non - detection of radio emission from snr in ngc 5253 supports the presence of very young nuclear starbursts ( less than 1 - 2 10@xmath23 yr , beck et al .
does the presence of wc stars lead to the high excitation of the gas , and thus the nebular emission as suggested by schaerer ( 1996 ) ? at the age of region _ a _ ( estimated from its large value of w ( ) ) the models of schaerer ( 1996 ) predict an important _ nebular _ contribution to the total @xmath34686 emission due to the large fraction of wc stars in the burst . given the weakness of the 4686 line in this region , its width is quite uncertain .
the presence of strong nebular lines of [ ] @xmath34658 and [ ] @xmath134711,4740 indicating high excitation resembles closely the cases of pox 4 ( cf .
kunth & sargent 1981 ; vc92 ) and mkr 1271 ( contini 1996 ) where the distinction between broad stellar and nebular is unclear .
the observations of region _ a _ are thus compatible with wc stars as the origin of nebular emission . at the age corresponding to region _
b _ the models of schaerer ( 1996 ) predict only broad emission , in agreement with the observations . compared to _ a _ , the weakness of the forbidden lines also indicates a lower excitation . _
the ionizing clusters : _ as pointed out by ksrwr97 the brightest uv cluster ( their uv1 which is included in our region _ b _ ) does not provide enough ionizing flux to explain the surface brightness of ngc 5253 corresponding to @xmath24 photons s@xmath25 ( martin & kennicutt 1995 ) derived assuming a constant extinction of @xmath11 0.47 . our observations , however , clearly reveal that region _ a _ ( corresponding to the optical maximum and also approximately to the emission peak ) contains a larger number of o stars than region _
b_. from we obtain an ionizing lyman continuum flux of @xmath26 and @xmath27 photon s@xmath25 for regions _ a _ and _ b _ respectively .
a _ is of higher extinction and thus clearly dominates the production of ionizing photons as compared to the brightest region in the uv . in view of the possible underestimation of extinction ( cf .
beck et al . 1996 ) and the strong spatial extinction variations ( ksrwr97 , calzetti et al .
1997 ) we suspect that the close area surrounding region _
( possibly including also _ h _ ) may well produce enough ionizing photons to explain the total observed surface brightness .
_ wr stars and the chemical enrichment : _ local overabundance of nitrogen in several regions of ngc 5253 is well established ( wr87 , wr89 , ksrwr97 ) .
the recent hst observations of ksrwr97 show n enrichment in two locations ( called hii-1 and hii-2 ) close to the peak of emission (= our region _ a _ ) , while n / o in their region uv1 ( included in our region _ b _ ) seems to be typical for metal - poor galaxies . assuming that wr stars eject a significant amount of nitrogen , our likely detection of wr stars in _ a _ provides a plausible source of n in the very proximity of the n - enriched regions .
region _ a _ , overlooked by ksrwr97 due to its low uv brightness , is thus most likely the `` hidden '' cluster containing the sources of the observed pollution .
more generally , the detection of wr stars of both wn and wc subtypes in both of our regions raises several questions .
the earlier finding of a possible carbon overabundance by wr89 , which originally lead these authors to suggest the presence of wc stars , was not confirmed by the results of ksrwr97 .
if the overabundance of n ( in the regions close to _ a _ ) is due to wn stars , why are the ejecta of wc stars ( presumably mostly he , c and o ) not detected ?
furthermore , why does region _
b _ ( = uv1 ) , which also harbours wr stars , not show any significant overabundance ?
abundance differences between these two regions might be related to their age difference : maybe mixing processes in _
b _ had time to dilute the ejecta with the ambient medium , whereas the younger region _ a _ is currently in a phase of heavy pollution .
hopefully a study of ngc 5253 will not only tell us its history of star formation but also more about still poorly understood nucleosynthetic yields in massive stars and mixing processes in the ism .
ds thanks daniela calzetti , rosa gonzlez - delgado , claus leitherer , gerhard meurer , and bill vacca for stimulating and useful discussions .
dk thanks eline tolstoy for a careful reading of the manuscript .
daniela calzetti also kindly provided us with hst images of ngc 5253 .
ds acknowledges a fellowship from the swiss national foundation of scientific research and partial support from the directors discretionary research fund of the stsci .
beck , s.c . ,
turner , j.l .
, ho , p.t.p . , lacy , j.h .
, kelly , d.m . , 1996 , , 457 , 610 burnstein , d. , heiles , c. , 1984 , , 54 , 33 campbell , a. , terlevich , r. , melnick , j. , 1986 , , 223 , 811 calzetti , d. , et al .
, 1997 , in preparation conti , p. , 1991
, , 377 , 115 conti , p. , vacca , w.d.w . , 1994 , , 423 , l97 contini , t. , 1996 , in `` wr stars in the framework of stellar evolution '' , 33rd lige int .
, eds . j.m .
vreux et al . ,
esteban , c. , peimbert , m. , 1995 , , 300 , 78 gorjian , v. , 1996 , , 112 , 1886 izotov , y.i .
, thuan , t.x . ,
lipovetsky , v.a . , 1994 , , 435 , 647 izotov , y.i . ,
thuan , t.x . ,
lipovetsky , v.a . , 1997 , , in press kingsburgh , r.l . , barlow , m.j . ,
storey , p.j .
, 1995 , , 295 , 75 kobulnicky , h.a . ,
skillman , e.d . ,
roy , j.r . ,
walsh , j.r . ,
rosa , m.r . , 1997 , , in press ( ksrwr97 ) kunth , d. , sargent , w.l.w .
, 1981 , , 101 , l5 maeder , a. , 1983 , , 120 , 113 martin , c.l . ,
kennicutt , r.c . , 1995 , , 447 , 171 meurer , g. , heckman , t.m . , leitherer , c. , kinney , a. , robert , c. , garnett , d.r . , 1995 , , 110 , 2665 meynet , g. , 1995 , , 298 , 767 pagel , b.e.j . , simonson , e.a . ,
terlevich r.j . ,
edmunds , m.g . , 1992 , , 255 , 325 sandage , a. , saha , a. , labhardt , l. , schwengeler , h. , panagia , n. , macchetto , f.d . , 1994 , , 423 , l13 schaerer , d. , 1996 , , 467 , l17 schaerer , d. , vacca , w.d.w . , 1996 , in `` wr stars in the framework of stellar evolution '' , 33rd lige int
. astroph .
, eds . j.m .
vreux et al . , p. 641
seaton , m.j .
, 1979 , , 187 , 73p smith , l.f . ,
shara , m.m . ,
moffat , a.f.j . , 1990 , , 358 , 229 smith , l.f . ,
shara , m.m . ,
moffat , a.f.j . , 1996 , , 281 , 163 vacca , w.d . , conti , p. , 1992
, , 401 , 543 walsh , j.r . ,
roy , j - r .
, 1987 , , 319 , l57 ( wr87 ) walsh , j.r . ,
roy , j - r .
, 1989 , , 239 , 297 ( wr89 ) lrrrrrrr hei & 4471 & 45.8 & 52.1 & 36.0 & 38.4 & 55.2 & 69.8 niii / ciii & 4645 & ( 6.2 ) & ( 6.7 ) & ( 39.3 ) & ( 40.7 ) & ( 35.4 ) & ( 40.1 ) [ feiii ] & 4658 & 13.0 & 13.8 & 19.1 & 19.8 & 22.7 & 25.5 heii & 4686 & 8.3 & 8.8 & 54.8 & 56.3 & 150.3 & 166.3 [ ariv ] & 4711 & 16.1 & 16.9 & ( 21.9 ) & ( 22.3 ) & ( 48.3 ) & ( 52.6 ) [ ariv ] & 4740 & 10.1 & 10.5 & ( 10.8 ) & ( 11.1 ) & ( 12.7 ) & ( 13.6 ) hei & 4922 & 10.0 & 9.8 & 10.9 & 10.7 & 10.1 & 9.7 [ oiii ] & 4959 & 2123.0 & 2059.7 & 1564.1 & 1538.8 & 1779.4 & 1685.1 [ oiii ] & 5007 & 6375.6 & 6103.1 & 4695.2 & 4589.0 & 5322.4 & 4917.1 [ ni ] & 5199 & 7.4 & 6.7 & 7.4 & 7.0 & & [ feiii ] & 5271 & 4.8 & 4.2 & 6.4 & 6.0 & & [ cliii ] & 5518 & 4.0 & 3.3 & 4.6 & 4.1 & & [ cliii ] & 5538 & 2.9 & 2.4 & & & & [ nii ] & 5755 & 5.4 & 4.3 & ( 2.8 ) & ( 2.4 ) & & civ & 5808 & 7.2 & 5.6 & 33.0 & 29.0 & & hei & 5876 & 142.2 & 109.5 & 127.6 & 111.0 & 126.9 & 79.2 [ oi ] & 6300 & 26.8 & 19.0 & 34.7 & 28.8 & 33.5 & 17.9 [ siii ] & 6312 & 40.2 & 28.4 & 24.7 & 20.6 & 22.1 & 11.8 [ oi ] & 6364 & 8.8 & 6.2 & 10.5 & 8.7 & 10.0 & 5.2 & & & & & & & @xmath10 & & 0.44 & & 0.20 & & 0.85 & @xmath28 & & 294 & & 109 & & 87 & @xmath29 & & 70.6 & 229.8 & 29.8 & 55.8 & 13.1 & 109.6
|
we present spectroscopic observations of the central star clusters in ngc 5253 the aim of which is to search for wc stars . our observations show the presence of wolf - rayet ( wr ) stars not only of wn but also of wc subtype in two star forming regions corresponding to the maximum optical and uv emission .
the massive star population we derive is consistent with young bursts of @xmath0 3 and 4 myr .
the region of maximum optical emission is found to provide the dominant contribution of the ionizing flux as opposed to the less extinguished region of maximum uv brightness .
the presence of wr stars near the n - enriched regions found by walsh & roy ( 1987 , 1989 ) and kobulnicky et al .
( 1997 ) suggests they are a possible source of n. it is presently unclear whether or not our detection of wc stars is compatible with the normal observed he / o and c / o abundance ratios .
_ apj letter , in press ( received : 2 november 1996 , accepted : 13 march 1997 ) _
# 1#2@xmath1 2c16o@xmath2
|
dimension reduction plays an important role in high dimensional data analysis .
one then wishes to reduce the dimension of a @xmath0-variate random vector @xmath1 using a transformation @xmath2 where the transformation matrix @xmath3 is a @xmath4 matrix with linearly independent columns , @xmath5 .
the column vectors of @xmath3 then span the @xmath6-dimensional subspace of interest .
the transformation to the subspace can also be done using the corresponding @xmath7 orthogonal projector @xmath8 .
the transformation @xmath9 projects the observations to a linear @xmath6-variate subspace .
there are two main types of dimension reduction unsupervised and supervised .
principal component analysis ( pca ) is perhaps the best known unsupervised dimension reduction method .
pca finds an orthogonal transformation matrix in such a way that the components in the new coordinate system are uncorrelated and ordered according to their variances . in dimension reduction ,
only the @xmath6 components with highest variances are taken .
independent component analysis ( ica ) is another example of an unsupervised dimension reduction method .
the fourth - order blind identification ( fobi ) @xcite procedure then finds a transformation matrix in such a way that the new components are uncorrelated with respect to two distinct scatter matrices , the regular covariance matrix and a scatter matrix based on fourth moments , and ordered according to their kurtosis values .
then the @xmath6 components with kurtosis values most deviating from that of a normal distribution are most interesting . in dimension reduction , the goal is often to use the transformed ( reduced ) variables to predict the value of a known response variable @xmath10 . in supervised dimension reduction , the joint distribution of @xmath11 and @xmath10
is then taken into consideration in the dimension reduction of @xmath11 , and it is hoped that @xmath12 . sliced inverse regression ( sir )
@xcite is a well known supervised dimension reduction method .
it is , along with ica , based on the comparison of two scatter matrices , where the second scatter matrix depends - unlike in unsupervised dimension reduction methods - on the joint distribution of @xmath11 and @xmath10 .
other well known dimension reduction methods are for example the sliced average variance estimate ( save ) and the principal hessian directions ( phd ) ( see * ? ? ?
* ; * ? ? ?
* respectively ) .
a recent contribution to the field of supervised dimension reduction methods is the supervised invariant coordinate selection ( sics ) @xcite .
it is an extension of invariant coordinate selection ( ics ) @xcite .
both ics and sics are based on the comparison of two scatter matrices @xmath13 and @xmath14 , and the transformation matrix is based on the eigenvalues and eigenvectors of @xmath15 . in sics however , the second scatter matrix depends on the joint distribution of @xmath11 and @xmath10 .
individual dimension reduction methods are usually adequate to find only special types of subspaces or special relationships between @xmath11 and @xmath10 .
for example , sir works well for linear relationships but may completely fail for other type of dependencies ( see for example * ? ? ? * ) .
hence , there is a need for new approaches , which would use the good properties of individual dimension reduction methods and combine the information in an efficient way .
the aim of this paper is to combine different dimension reduction methods in such a way that the best qualities " of each method are picked up .
to combine different dimension reduction methods is to say that we combine the individual orthogonal projectors possibly with various ranks and find an average orthogonal projector " ( aop ) with an optimized rank .
our approach is similar to the approach in @xcite .
the idea is to find the aop which is , on the average , closest to individual orthogonal projectors with respect to some distance criterium .
the paper is organised as follows . in section [ sec : theory ] we discuss subspaces and propose a generalization of the crone and crosby distance . @xcite
considered subspaces of equal dimensions , whereas our weighted distance allows subspaces of different dimensions .
some natural choices of weights are given .
furthermore , the concept of averages of subspaces is discussed . in section [ sec : applications ] the performance of the weighted distance and the different aop s is evaluated in two unsupervised dimension reduction applications and one supervised dimension reduction simulation study . the paper ends with some final remarks .
we first consider linear subspaces in @xmath16 with a fixed dimension @xmath6 , @xmath17 . a linear subspace and the distances between the subspaces can be defined in several ways . 1 .
the subspace is defined as a linear subspace spanned by the linearly independent columns of a @xmath18 matrix @xmath3 , that is , @xmath19 this definition based on a matrix @xmath3 is a bit obscure in the sense that @xmath20 for all full - rank @xmath21 matrices @xmath22 . according to this definition , the same subspace can in fact
be fixed by any member in a set of matrices equivalent to @xmath3 , @xmath23 the non - uniqueness of @xmath3 may cause technical problems in the estimation of a subspace . consider two @xmath4 matrices @xmath24 and @xmath25 with rank @xmath6 .
then a measure of distance between subspaces spanned by @xmath24 and @xmath25 can be defined as @xmath26 where @xmath27 are the squared canonical correlations between @xmath24 and @xmath25 @xcite and @xmath28 .
note that if @xmath24 and @xmath25 are equivalent then the squared canonical correlations are all 1 .
the subspace is defined as a linear subspace spanned by the orthonormal columns of a @xmath18 matrix @xmath29 .
note that , starting with @xmath3 , one can choose @xmath30 for this second definition .
unfortunately , the definition is still obscure as @xmath31 for all orthonormal @xmath21 matrices @xmath32 , and the same subspace is given by any matrix in the class of equivalent orthonormal matrices @xmath33 the principal angles @xmath34 $ ] between the subspaces @xmath35 and @xmath36 with corresponding @xmath6-variate direction vectors @xmath37 and @xmath38 @xmath39 , are recursively defined by maximizing @xmath40 subject to the constraints @xmath41 , and @xmath42 , @xmath43 .
the @xmath44th principal angle is then @xmath45 , @xmath46 , and a measure of distance between the subspaces may be obtained as @xmath47 .
it is easy to see that it equals to @xmath48 .
the subspace is defined as the linear subspace given by an orthogonal projector @xmath49 , that is , a @xmath50 transformation matrix @xmath49 such that @xmath51 matrix @xmath49 provides a unique way to fix the subspace @xmath52 .
note that , starting from @xmath3 , one can define @xmath53 in a unique way .
starting from @xmath54 gives similarly @xmath55 .
the squared distance between the subspaces given by two orthogonal projectors @xmath56 and @xmath57 may then be defined as the matrix ( frobenius ) norm @xmath58 @xcite use @xmath59 as a distance between two @xmath6-dimensional subspaces of @xmath60 given by orthogonal projectors @xmath56 and @xmath57 .
it is then easy to see that @xmath61 and that the distance obeys the triangular inequality @xmath62 for any orthogonal projectors @xmath56 , @xmath57 and @xmath63 .
assume next that the ranks of the projection matrices @xmath56 and @xmath57 are @xmath64 and @xmath65 , respectively , where @xmath66 .
for completeness of the theory , we also accept projection matrices @xmath67 with rank @xmath68 .
as @xmath69 , one possible extension of the above distance is @xmath70^{1/2}. $ ] then @xmath71 but , unfortunately , the triangular inequality is not true for this distance .
we therefore consider other extensions of the metric by @xcite .
let @xmath72 be positive weights attached to dimensions @xmath73 .
( we will later see that the choice of @xmath74 is irrelevant for the theory . )
we then give the following definition . a weighted distance between subspaces @xmath56 and
@xmath57 with ranks @xmath64 and @xmath65 is given by @xmath75 the weights are used to make the orthogonal projectors @xmath56 and @xmath57 with different ranks more comparable in some sense . as the distance @xmath76 is based on the matrix ( frobenius ) norm , ( i ) @xmath77 , ( ii ) @xmath78 if and only if @xmath79 , ( iii ) @xmath80 , and ( iv ) @xmath81 , and we have the following result .
[ metricwithweights ] for all weight functions @xmath82 , @xmath76 is a metric in the space of orthogonal projectors , and the strict lower and upper bounds of @xmath83 for the dimensions @xmath64 and @xmath65 are @xmath84 where @xmath85 some interesting choices of the weights are , for @xmath86 , @xmath87 weights in ( @xmath88 ) give the distance by @xcite .
weights in ( @xmath89 ) and ( @xmath90 ) standardize the matrices so that @xmath91 and @xmath92 , respectively , if @xmath93 .
it is remarkable that @xmath94 where @xmath95 is a correlation between vectorized @xmath56 and @xmath57 .
proposition [ metricwithweights ] implies that , for nonzero @xmath64 and @xmath65 , the distances @xmath83 get any values on the closed intervals @xmath96 , \\ & ( b ) : & \left[\frac 12 \big|k_1^{-1}-{k_2}^{-1}\big| , \frac 12\left({k_1}^{-1}+{k_2}^{-1 } \right)+{k_1^{-1}k_2^{-1 } } \min\ { p - k_1-k_2,0\}\right],\ \ \mbox{and}\\ & ( c ) : & \left [ 1-\min\{k_1^{1/2}k_2^{-1/2},k_1^{-1/2}k_2^{1/2 } \ } , 1 + k_1^{-1/2 } k_2^{-1/2}\min\ { p - k_1-k_2,0\}\right].\end{aligned}\ ] ] if @xmath97 , for example , then @xmath83 is simply @xmath98 .
recall that , for all three choices of weights , the distance is zero only if @xmath79 ( and @xmath99 ) . for weights @xmath100 ,
the largest possible value for @xmath101 is @xmath102 and it is obtained if and only if @xmath56 and @xmath57 are orthogonal and @xmath103 ( i.e. , @xmath104 ) . for weights
@xmath105 , @xmath106 , and @xmath107 if and only if @xmath56 and @xmath57 are orthogonal and @xmath108 .
finally , for weights @xmath109 , the maximum value @xmath110 for @xmath111 is attained as soon as @xmath56 and @xmath57 are orthogonal and @xmath112 .
the following two special cases illustrate the differences between the three distances . 1 .
first , consider the case when @xmath113 .
then naturally @xmath114 and @xmath115 and therefore , for @xmath116 and with @xmath117 , @xmath118 one can see that @xmath119 depends only on the ratio between @xmath64 and @xmath65 , which can be seen as a nice feature . @xmath120 and
@xmath121 however depend additionally on the actual values of @xmath64 and @xmath65 .
2 . second , consider the case when @xmath122 and @xmath123 are orthogonal , that is , when @xmath124 .
then @xmath125 and therefore , for nonzero @xmath64 and @xmath65 , @xmath126 it is natural to think subspaces that are orthogonal to each other are furthest apart possible .
this information is apparent in @xmath119 .
however , interpreting both @xmath120 and @xmath121 is again more difficult since they depend on the actual values of @xmath64 and @xmath65 .
consider orthogonal projectors @xmath127 with ranks @xmath128 . to combine the projectors we give the following [ avedef ] the average orthogonal projector ( aop ) @xmath129
based on weights @xmath72 is an orthogonal projector that minimizes the objective function @xmath130 to find the aop , we can use the following result .
[ aoplemma ] the aop @xmath131 maximizes the function @xmath132 where @xmath133 is a regular average of weighted projectors , and @xmath6 is the rank of @xmath49 . naturally , @xmath134 is symmetric and non - negative definite , but not a projector anymore . in the following derivations ,
we need its eigenvector and eigenvalue decomposition @xmath135 where @xmath136 and @xmath37 is the eigenvector corresponding to the eigenvalue @xmath137 .
recall that the eigenvectors are uniquely defined only for eigenvalues that are distinct from other eigenvalues . using the lemma [ aoplemma ] and the eigenvector and eigenvalue decomposition @xmath134 , our main result easily follows .
the rank @xmath6 of the aop @xmath129 maximizes the function @xmath138 where @xmath139 are the eigenvalues of @xmath134 .
moreover , @xmath140 where @xmath141 are the eigenvalues corresponding to eigenvalues @xmath142 .
note that the calculation of the aop @xmath131 is easy , only the eigenvalues and eigenvectors of @xmath134 are needed .
the aop @xmath131 is not always unique .
this happens for example if the rank of an aop is @xmath6 and @xmath143 .
consider now the three weight functions @xmath144 the function @xmath145 for these three weight functions is , for @xmath86 , @xmath146 note that @xmath147 for all weights @xmath82 . to find local maxima for these functions
, one can then use the results @xmath148 for @xmath149 .
note that for @xmath150 , @xmath151 is a concave function and the global maximum is simply the largest @xmath6 with the eigenvalue @xmath152 .
the functions in @xmath153 and @xmath154 are not concave , however , and the global maximum is found by computing all the values @xmath151 , @xmath155 .
in this section we discuss the performance of the averages of orthogonal projectors ( aop ) for three different dimension reduction problems .
the orthogonal projectors and their combinations aim for different targets in different applications .
each problem with natural projectors will be first shortly introduced , and then the performance of aop is demonstrated using simulation studies . the computations in this section are done using r @xcite by mainly using the packages dr @xcite , mnm @xcite , pcapp @xcite and robustbase @xcite .
classical principal component analysis ( pca ) may be based on the eigenvector and eigenvalue decomposition of the covariance matrix of a @xmath0-variate random vector @xmath11 , that is , on @xmath156 where @xmath157 are the ordered eigenvalues and @xmath158 are the corresponding eigenvectors .
orthogonal projector @xmath159 then projects @xmath0-variate observations to the @xmath6-variate subset with maximum variation .
it is unique if @xmath160 .
let @xmath161 be the cumulative distribution function of @xmath11 .
a @xmath50 matrix valued functional @xmath162 is a scatter matrix if @xmath162 is a non - negative definite and symmetric matrix with the affine equivariance property @xmath163 it is remarkable that , if @xmath11 has an elliptic distribution then the ordered eigenvectors of @xmath164 are those of @xmath165 . therefore ,
in the elliptic case , any scatter matrix can be used to find @xmath166 and the matrix @xmath49 is a well - defined population quantity even if the second moments ( and the covariance matrix ) do not exist .
naturally , the sample statistics corresponding to different scatter matrices then have different statistical ( efficiency and robustness ) properties . for a fixed value of @xmath6
, one can then try to `` average '' these different pca approaches to get a compromise estimate .
we next illustrate the performance of the aop in the following simple scenario .
let first @xmath167 @xmath168 .
we choose @xmath169 and wish to estimate @xmath170 .
let then @xmath171 be a random sample from @xmath167 , and find an estimate @xmath172 , an orthogonal projector with rank @xmath169 obtained from the sample covariance matrix .
this estimate is then combined with three robust estimates , namely , @xmath173 : : that is based on tyler s shape matrix @xcite with the affine equivariant version of spatial median as a multivariate location estimate @xcite . @xmath174 : : that is based on the minimum covariance determinant ( mcd ) estimator @xcite . @xmath175 : : that is based on projection pursuit ( pp ) approach for pca with the median absolute deviation ( mad ) criterion as suggested in @xcite . in the simulations
, @xmath171 was a random sample from @xmath176 with @xmath177 , and the sampling was repeated 1000 times .
as @xmath178 is fixed , we use only @xmath100 as the weight function .
the average squared distances @xmath179 between the four projector estimates , their aop @xmath180 , and @xmath181 are shown in table [ tab : distmatrix ] .
a similar simulation study was conducted but with observations coming from a heavy - tailed elliptical @xmath182 distribution with @xmath183 as proportional eigenvalues .
note that the regular scatter matrix does not exist in this case but the true projection matrix is still well defined .
.average squared distances @xmath184 between the four projector estimates , their aop @xmath180 , and true @xmath181 .
for all projectors , rank @xmath169 .
the averages are based on 1000 random samples of size @xmath177 from @xmath176 . [ cols="<,<,<,<,<,<,<",options="header " , ] in the previous section we used projection pursuit ( pp ) approach for principal component analysis .
pp is a much more general technique , however , and there are many other types of indices than just measures of variation to define `` interesting '' one - dimensional directions .
pp actually dates back to @xcite and usually one searches for nongaussian directions . for
a recent review of the existing indices , see for example @xcite .
a major challenge in pp is that it is computationally difficult to find the direction which globally maximizes the index and that there are usually several local maxima .
however , since the local maxima may be also of interest , one possible strategy , as detailed in @xcite , is to run the algorithm many times using different initializations . with this strategy ,
the user has many projectors of rank one but many of them are usually redundant .
so , it is of particular interest to summarize all these projectors in order to extract the directions that are useful and unique .
it means that , in that case , one is interested in an average projector of projectors with rank one that may have a higher rank . to demonstrate the interest of aop in the context of pp
, we choose the deflation - based fastica method @xcite as an example since it is well - understood and computationally quite simple .
while deflation - based fastica is originally developed in the context of independent component analysis ( ica ) , it can be seen as a traditional pp approach when only one direction is extracted . for a random variable @xmath11 with the standardized version @xmath185
, deflation - based fastica maximizes a measure of non - gaussianity of the form @xmath186 , under the constraint that @xmath187 , where @xmath188 is a selected twice differentiable nonlinear nonquadratic function with @xmath189 .
the final pp direction is then @xmath190 , and the corresponding orthogonal projector is @xmath191 . in our simulations , we use four common choices of @xmath192 with derivative functions @xmath193 : ( i ) @xmath194 , ( ii ) @xmath195 , ( iii ) @xmath196 , and ( iv ) @xmath197 . if there are more than one non - gaussian direction in the data , the direction to be found depends heavily on the initial value of the algorithm , see e.g. @xcite . in our simulation study , we choose a 10-variate @xmath198 where the first three variables are mixtures of gaussian distributions and @xmath199 , for @xmath200 . more precisely ,
@xmath201 with @xmath202 , @xmath203 and @xmath204 ; @xmath205 with @xmath206 , @xmath207 and @xmath208 , and @xmath209 with @xmath210 , @xmath211 and @xmath212 .
we generated 1000 random samples of sizes @xmath213 from the 10-variate distribution described above . for each sample , we found 100 one - dimensional pp directions ( 4 choices of @xmath188 , 25 random initial values for the algorithm for each choice of @xmath188 ) .
for each sample , 100 pp projectors were then averaged using each of the three weight functions @xmath100 , @xmath105 and @xmath109 . the average projector should in this setting then be close to the projector @xmath214 with rank 3 that picks the three non - gaussian components of the data .
figure [ fig : rank_freq ] shows the relative frequencies of the ranks of the aops obtained with the three weight functions in 1000 repetitions .
clearly the weight function @xmath100 is not appropriate in this application because @xmath215 implies that @xmath216 with @xmath217 , which means that there can not be more than one eigenvalue larger than 1/2 and , consequently , the rank @xmath6 equals zero or one . with weight functions @xmath105 and @xmath109 , the correct rank 3 is obtained in 82.6% and 68.3% of the runs , respectively .
it is also hoped that the aops are close to the true projector @xmath218 . to evaluate that
, we found that the the average distances @xmath179 between @xmath219 and @xmath218 , between @xmath220 and @xmath218 , and between @xmath221 and @xmath218 were 1.005 , 0.122 , and 0.199 , respectively .
the same numbers for the distances based on @xmath105 and @xmath109 are 0.335 , 0.018 , and 0.035 , and 0.425 , 0.043 , and 0.0736 , respectively .
notice that , for all distances , the aop @xmath220 is closest on average to the true value @xmath218 . in the pca application
, we used the same @xmath169 for orthogonal projectors and their aop . in the pp application ,
the rank of the orthogonal projectors was taken as one while the rank of their aop was not fixed .
however , for many dimension reduction methods , the ranks of the individual orthogonal projectors are not fixed but also estimated from the data , and the ranks may differ from one method to another .
we now look at this scenario in the framework of supervised dimension reduction . in supervised dimension reduction ,
one often assumes that a response variable @xmath10 and the @xmath0-vector @xmath11 are related through @xmath222 with an unknown function @xmath223 and an unknown error variable @xmath224 .
the goal of supervised dimension reduction is to estimate the value of @xmath6 and the matrix @xmath225 to obtain @xmath226 with rank @xmath6 .
hence , for supervised dimension reduction , the joint distribution of @xmath10 and @xmath11 is of interest and , for the matrix @xmath3 , it holds that @xmath227 .
many supervised dimension reductions are suggested in the literature and their performances often strongly depend on the unknown function @xmath223 .
the well - known sliced inverse regression ( sir ) for example may not find directions with nonlinear dependencies while , on the other hand , principal hessian directions ( phd ) can not find linear relationships .
hence , when using supervised dimension reduction methods in practice , the estimated rank @xmath6 and the corresponding projector might differ considerably depending on the method .
we propose to use the aop in order to summarize in an efficient way the information brought by the complementary estimation strategies .
in our example we generate data sets from the following three models .
@xmath228 where @xmath229 , @xmath230 , @xmath231 and @xmath232 s are all 10-dimensional row vectors @xmath233 hence @xmath234 for model m1 and @xmath235 for models m2-m3 . in each case
, we generated 100 samples of size 400 .
in our illustration , we use supervised dimension reduction methods implemented in the dr package that provide both the estimate of @xmath6 and the orthogonal projector estimate with the same rank @xmath6 .
the estimation strategies are then ( i ) sliced inverse regression ( sir ) , ( ii ) sliced variance estimation ( save ) , ( iii ) inverse regression estimation ( ire ) , and three types of principal hessian directions ( phd ) , namely , ( iv ) response based principal hessian directions ( phdy ) , ( v ) residual based principal hessian directions ( phdr ) , and ( vi ) the so called @xmath236-based principal hessian directions ( phdq ) . for details about these estimation methods , see @xcite and references therein .
we also add here ( vii ) pca with @xmath6 chosen simply as the number of eigenvalues larger than 1 .
naturally , pca ignores @xmath10 and is therefore not supervised .
( its use could be motivated by the aim to avoid directions with small variation . in our case
it just provides random projectors . ) in the following we want to compare the seven methods above and their aops based on the weight functions @xmath105 and @xmath109 .
the use of @xmath100 is not reasonable with varying @xmath6 .
we consider here the following four aops . 1 .
the aop using @xmath105 with fixed and true @xmath6 .
2 . the aop using @xmath109 with fixed and true @xmath6 .
the aop using @xmath105 with estimated @xmath6 .
4 . the aop using @xmath109 with estimated @xmath6 .
some simulation results are collected in figures [ fig : avproj_distances_m2]-
[ fig : avproj_distances_m6 ] .
the figures show the boxplots for the observed @xmath237 and @xmath238 distances between the true orthogonal projector and the projector estimates coming from different dimension reduction approaches .
consider first the behavior of the estimates in the model m1 with @xmath234 ( see figure [ fig : avproj_distances_m2 ] ) .
the performances of sir , save and phd estimates seem to be very similar and they usually find only one direction .
( for example , sir finds only the component with linear dependence , and save only the component of quadratic dependence . )
the same seems to be true with ire but with more varying estimates .
recall that , if @xmath113 and @xmath239 and @xmath240 , then @xmath241 , and the average distances of sir , save and phd estimates tend to be close to @xmath242 respectively .
the aop estimates then nicely pick up the two dimensions and clearly outperform other estimates .
note that there is no big difference between aop estimates with known @xmath6 and aop estimates with estimated @xmath6 .
the aop estimate based on @xmath109 seems to be better .
pca has a poor performance as expected .
figure [ fig : avproj_distances_m4 ] shows the results when the observations come from the model m2 .
the model with linear dependence only is then of course the model where sir is the best one .
ire also performs quite well but , for most samples , save and phd approaches do not find any solution at all . recall that , if @xmath97 and @xmath243 then @xmath244 .
the aop estimates seem often to pick up the correct subspace , and there is no real difference between the @xmath105 and @xmath109 estimates .
this time , the aop estimates with known dimension @xmath235 seem to perform better than the aom estimates with estimated @xmath6 .
figure [ fig : avproj_distances_m6 ] gives the results for the model m3 with @xmath235 and quadratic dependence .
save and phd approaches work very well , and sir and ire completely fail in this case .
again , all aop estimates neglect the bad estimates and pick up nicely the correct one direction . as in the other cases , pca provides a random reference method with a bad performance indeed .
dimension reduction and subspace estimation is a topic with increasing relevance since modern datasets become larger and larger .
different approaches have different shortcomings and combining the results coming from different approaches might give a better total overview . in this paper , we propose a generalization of the crone and crosby distance for the orthogonal projectors , a weighted distance that allows to combine subspaces of different dimensions .
some natural choices of weights are considered in detail .
the performance of three weighted distances and the combining approach is illustrated via simulations which show that each of them has its own justification depending on the problem at hand .
similar to other areas of statistics , this kind of `` model averaging '' seems to be a way to combine information from competing estimates and to give a better idea of the true model at hand .
99 natexlab#1#1 cardoso , j. f. ( 1989 ) .
source separation using higher order moments .
_ proceedings of ieee international conference on acoustics , speech and signal processing _ , 4 , 21092112 .
cook , r. d. and weisberg , s. ( 1991 ) . sliced inverse regression for dimension reduction : comment .
_ journal of the american statistical association _ , 86 , 328332 .
crone , l. j. and crosby , d. s. ( 1995 ) .
statistical applications of a metric on subspaces to satellite meteorology . _ technometrics _ , 37 , 324328 .
croux , c. , and ruiz - gazen , a. ( 2005 ) .
high breakdown estimators for principal components : the projection - pursuit approach revisited .
_ journal of multivariate analysis _ , 95 , 206226 .
filzmoser , p. , fritz , h. , and kalcher , k. ( 2012 ) .
pcapp : robust pca by projection pursuit .
r package version 1.9 - 47 .
friedman , j. h. and tukey , j.w .
( 1974 ) . a projection pursuit algorithm for exploratory data analysis . _ ieee trans .
computers c _ , 23 , 881889 .
hettmansperger , t. p. , and randles , r. h. ( 2002 ) .
a practical affine equivariant multivariate median .
_ biometrika _ , 89 , 851860 .
hotelling , h. ( 1936 ) .
relations between two sets of variates .
_ biometrika _ , 28 , 321377 .
hyvrinen , a. ( 1999 ) .
fast and robust fixed - point algorithms for independent component analysis . _ ieee transactions on neural networks _ , 10 , 626634 .
li , k .- c .
( 1991 ) . sliced inverse regression for dimension reduction .
_ journal of the american statistical association _ , 86 , 316327 .
li , k .- c .
( 1992 ) . on principal hessian directions for data visualization and dimension reduction : another application of stein s lemma .
_ journal of the american statistical association _ , 87 , 10251039 .
liski , e. , nordhausen , k. and oja , h. ( 2012 ) .
supervised invariant coordinate selection .
submitted .
nordhausen , k. , ilmonen , p. , mandal , a. , oja , h. and ollila , e. ( 2011 ) .
deflation - based fastica reloaded .
_ proceedings of 19th european signal processing conference 2011 ( eusipco 2011 ) _ , 18541858 .
nordhausen , k. and oja , h. ( 2011 ) .
multivariate l1 methods : the package mnm .
_ journal of statistical software _ , 43 , 128 .
r development core team ( 2012 ) .
r : a language and environment for statistical computing .
r foundation for statistical computing .
vienna , austria .
rodriguez - martinez , e. , goulermas , j.y . ,
mu , t. , and ralph , j.f .
automatic induction of projection pursuit indices .
_ ieee transactions on neural networks _ , 21 , 12811295 .
rousseeuw , p. ( 1986 ) .
multivariate estimation with high breakdown point . in _ mathematical statistics and applications _
grossman , w.,pflug , g. , vincze , i. , and wertz , w. ) , 283297 .
dordrecht , reidel .
rousseeuw , p. , croux , c. , todorov , v. , ruckstuhl , a. , salibian - barrera , m. , verbeke , t. , koller , m. , and maechler , m. ( 2012 ) .
robustbase : basic robust statistics .
r package version 0.9 - 2 .
ruiz - gazen , a. , berro , a. and larabi marie - sainte , s. ( 2010 ) .
detecting multivariate outliers using projection pursuit with particle swarm optimization .
_ compstat2010 _ , 8998 .
tyler , d. e. , ( 1987 ) .
a distribution - free m - estimator of multivariate scatter . _ the annals of statistics _ , 15 , 234251 .
tyler , d. e. , critchley , f. , dmbgen , l. , and oja , h. ( 2009 ) .
invariant co - ordinate selection .
_ journal of the royal statistical society , series b _ , 71 , 549592 .
weisberg , s. ( 2002 ) .
dimension reduction regression in r. _ journal of statistical software _
, 7 , 122 .
[ auxiliarylemma ] for two @xmath50 orthogonal projectors @xmath56 and @xmath57 with ranks @xmath64 and @xmath65 , respectively , @xmath245 * proof * first note that @xmath246 and @xmath247 where @xmath35 has @xmath64 orthonormal columns and @xmath36 has @xmath65 orthonormal columns . then @xmath248 . as @xmath249 and
@xmath250 one can conclude that @xmath251 .
similarly , @xmath252 and therefore @xmath253 , and the result follows .
note also that the lower and upper bounds in the above lemma are fixed .
the upper bound is obtained with the choices @xmath254 and the lower bound with the choices @xmath255 where @xmath256 is a @xmath0-vector with the @xmath44th component one and other components zero .
* proof of proposition 2.2 * the aop @xmath131 maximizes @xmath262 where @xmath6 is the rank of @xmath49 .
assume first that @xmath86 is fixed and @xmath263 where @xmath32 has @xmath6 orthonormal columns .
then @xmath264 is maximized as soon as @xmath265 is maximized .
then , as @xmath266 , @xmath267 is maximized if @xmath268 , and the maximum value is @xmath269 . for fixed @xmath86 , the maximum value of @xmath264 is then @xmath270 , and @xmath271 .
the result follows .
|
dimensionality is a major concern in analyzing large data sets .
some well known dimension reduction methods are for example principal component analysis ( pca ) , invariant coordinate selection ( ics ) , sliced inverse regression ( sir ) , sliced average variance estimate ( save ) , principal hessian directions ( phd ) and inverse regression estimator ( ire ) .
however , these methods are usually adequate of finding only certain types of structures or dependencies within the data .
this calls the need to combine information coming from several different dimension reduction methods .
we propose a generalization of the crone and crosby distance , a weighted distance that allows to combine subspaces of different dimensions .
some natural choices of weights are considered in detail .
based on the weighted distance metric we discuss the concept of averages of subspaces as well to combine various dimension reduction methods .
the performance of the weighted distances and the combining approach is illustrated via simulations .
keywords : dimension reduction ; distance ; metric ; principal component analysis ; projection pursuit ; subspace .
|
physical systems undergo phase transitions when external parameters such as the temperature ( @xmath0 ) or a chemical potential ( @xmath4 ) are changed .
a phase diagram provides intrinsic knowledge on the structure of the matter under study . in other words , it tells us how matter organizes itself under external conditions at a given degrees of freedom .
the theory of strong interactions , qcd , predicts that nuclear matter at high temperature and/or density makes a transition from a state where quarks and gluons are confined and chiral symmetry is broken to a state where quarks and gluons are de - confined and chiral symmetry is restored @xcite .
qcd has several conserved quantities : baryon number ( @xmath5 ) , electric charge ( @xmath6 ) , and strangeness ( @xmath7 ) .
each of these is associated with a chemical potential . as a result ,
the qcd phase diagram is four - dimensional . @xmath8 and @xmath9 are relatively small compared to @xmath10 ( baryonic chemical potential ) in high energy nuclear collisions @xcite .
the @xmath0 and @xmath10 are varied in a typical qcd phase diagram as shown in fig .
[ fig1 ] @xcite . at high temperature and density
the phase is governed by quark and gluon degrees of freedom and is commonly referred to as the quark gluon plasma ( qgp ) @xcite . at large densities and low temperatures other interesting phases
related to neutron star @xcite and color superconductivity @xcite starts to appear .
these qcd transitions which occurred in the early universe have the right energy scale to be accessible by the experiments . the fig .
[ fig1 ] shows the parts of the phase diagram explored by several accelerator based experimental programs .
experimentally , this is done by varying the beam energy .
both , @xmath0 and @xmath11 , vary as the function of the center - of - mass energy ( @xmath12 ) @xcite .
this strategy is being followed by experimental programs at large hadron collider ( lhc ) at cern , relativistic heavy ion collider ( rhic ) at bnl and super proton synchrotron ( sps ) at cern , and will be followed at facility for antiproton and ion research ( fair ) at gsi and nuclotron - based ion collider facility ( nica ) at jinr . among these experiments ,
we discuss in section 2 , selected results from rhic which provides evidence for the formation of a qgp @xcite .
theoretically , finite temperature lattice qcd calculations at @xmath10 = 0 suggest a cross - over @xcite above a temperature , @xmath13 , of about 170 to 190 mev from a system with hadronic degrees of freedom to a qgp @xcite . at large @xmath10
, several qcd based calculations find the quark - hadron phase transition to be of first order @xcite .
going towards the smaller @xmath10 region , the point in the qcd phase plane ( @xmath0 vs. @xmath10 ) where the first order phase transition ends is the qcd critical point ( cp ) @xcite .
the focus in the coming decade would be on attempts to locate the cp both experimentally and theoretically .
current theoretical calculations are highly uncertain about the location of the cp .
this is primarily because the lattice qcd calculations at finite @xmath10 face numerical challenges .
the experimental plan ( discussed in section 3 ) is to vary the @xmath12 of heavy - ion collisions to scan the phase plane and , at each energy , search for signatures of the cp that might survive the evolution of the system . in the last section of the review
we discuss the thermodynamic properties of the hadronic phase .
finally we end with a summary of our current understanding of phase diagram and a brief outlook .
enhancement of strange hadron production in high energy heavy - ion collisions @xcite due to formation of qgp is one of the four classic signatures in this field .
the other three being , enhanced direct photon @xcite and dilepton @xcite production and suppression of j/@xmath14 production @xcite in heavy - ion collisions relative to @xmath15+@xmath15 collisions . in a qgp
, thermal @xmath16 and @xmath17 quarks can be produced by gluon - gluon interactions @xcite .
these interactions could occur very rapidly and the @xmath16-quark abundance would equilibrate . during hadronisation , the @xmath16 and @xmath17 quarks from the plasma coalesce to form @xmath18 mesons .
production by this process is not suppressed as per the ozi ( okubo - zweig - izuka ) rule @xcite .
this , coupled with large abundances of strange quarks in the plasma , may lead to a dramatic increase in the production of @xmath18 mesons and other strange hadrons relative to non - qgp @xmath15+@xmath15 collisions . , @xmath18 , @xmath19 and @xmath20 normalized to average number of participating nucleons ( @xmath21 ) in nucleus - nucleus collisions to corresponding yields in inelastic @xmath15+@xmath15 collisions as a function of @xmath21 at 200 gev .
lower panel : same as above for @xmath18 mesons in cu+cu collisions at 200 and 62.4 gev .
the @xmath15+@xmath15 collision data at 200 gev are from star experiment and at 62.4 gev from isr @xcite .
the error bars shown represent the statistical and systematic errors added in quadrature .
figure is taken from ref .
@xcite . ] such predictions of relative enhancement of strange hadron was challenged by an alternate idea of canonical suppression of strangeness in small systems being the source of strangeness enhancement in @xmath22 , @xmath23 and @xmath24 hadrons in high energy heavy - ion collisions @xcite .
the strangeness conservation laws require the production of an @xmath17-quark for each @xmath16-quark in the strong interactions .
the main argument in such canonical models is that the energy and space time extensions in smaller systems may not be sufficiently large .
this leads to a suppression of open - strange hadron production in small collision systems .
these statistical models fit the data reasonably well @xcite .
these models predict two things : ( a ) strangeness enhancement in nucleus - nucleus collisions , relative to @xmath15+@xmath15 collisions , should increase with the valence strange quark content of the hadrons and ( b ) the enhancement is predicted to decrease with increasing beam energy @xcite .
discriminating between the two scenarios , qgp versus the canonical suppression , using the experimental data on @xmath25 , @xmath22 , @xmath23 and @xmath24 hadrons has been , to some extent , ambiguous .
enhancement of @xmath26 production ( zero net - strangeness , hence not subjected to the canonical suppression ) in heavy - ion collisions relative to @xmath15+@xmath15 collisions would clearly indicate the formation of a gluon rich qgp in these collisions .
this would then rule out canonical suppression scenario .
figure [ phi ] shows the enhancement of strange hadron production at rhic @xcite .
upper panel shows the ratio of strange hadron production normalized to @xmath21 in nucleus - nucleus collisions relative to corresponding results from @xmath15+@xmath15 collisions at 200 gev .
the results are plotted as a function of @xmath21 .
@xmath27 , @xmath19 and @xmath20 are found to exhibit an enhancement ( value @xmath28 1 ) that increases with the number of strange valence quarks .
furthermore , the observed enhancement in these open - strange hadrons increases with collision centrality , reaching a maximum for the most central collisions .
however , the enhancement of @xmath18 meson production shows a deviation from the ordering in terms of the number of strange constituent quarks .
more explicitly , this enhancement is larger than for @xmath27 and @xmath19 , at the same time being smaller than in case of @xmath20 . despite being different particle types ( meson - baryon ) and having different masses ,
the results for @xmath27 and @xmath19 are very similar in the entire centrality region studied .
this rules out a baryon - meson effect as being the reason for the deviation of @xmath18 mesons from the number of strange quark ordering seen in fig .
[ phi ] ( upper panel ) .
the observed deviation is also not a mass effect as the enhancement in @xmath18 meson production is larger than that in @xmath19 ( which has mass close to that of the @xmath18 ) .
further in heavy - ion collisions , the production of @xmath18 mesons is not canonically suppressed due to its @xmath16@xmath17 structure .
the observed enhancement of @xmath18 meson production then is a clear indication for the formation of a dense partonic medium being responsible for the strangeness enhancement in au+au collisions at 200 gev . the observed enhancement in @xmath18 meson production being related to medium density is further supported by the energy dependence shown in the lower panel of fig . [ phi ] .
the @xmath18 meson production relative to @xmath15+@xmath15 collisions is larger at higher beam energy , a trend opposite to that predicted in canonical models for other strange hadrons .
in addition measurements have shown that @xmath18 meson production is not from the coalescence of @xmath29 and is minimally affected by re - scattering effects in the medium @xcite . measurements also indicate that @xmath18 mesons are formed from the coalescence of seemingly thermalized strange quarks @xcite .
all these observations put together indicate the formation of a dense partonic medium in heavy - ion collisions where strange quark production is enhanced .
this in turn suggests that the observed centrality dependence of the enhancement for other strange hadrons ( seen in fig .
[ phi ] ) is likely to be related to the same reasons as in the case of the @xmath18 meson , that it is due to the formation of a dense gluon rich partonic medium in the collisions .
these experimental data rule out the possibility of canonical suppression being the only source of the observed strangeness enhancement at @xmath30= 200 gev .
one of the most exciting results to date at rhic is the discovery of suppression in the production of high transverse momentum ( @xmath31 ) mesons in nucleus - nucleus collisions compared to corresponding data from the binary collision scaled p+p collisions @xcite .
this has been interpreted in terms of energy loss of partons in qgp .
this phenomena is called as the jet quenching in a dense partonic matter @xcite .
the energy loss by energetic partons traversing the dense medium formed in high - energy heavy - ion collisions is predicted to be proportional to both the initial gluon density @xcite and the lifetime of the dense matter @xcite .
the results on high-@xmath31 suppression are usually presented in terms of the nuclear modification factor ( @xmath32 ) , defined as : @xmath33 where the overlap integral @xmath34 with @xmath35 being the number of binary collisions commonly estimated from glauber model calculation @xcite . ) for mesons and direct photons as measured in rhic experiments at midrapidity for central au+au collisions at @xmath36 = 200 gev .
also shown are the @xmath37 for charged pions for @xmath36 = 200 gev .
the lines are results from various model calculations .
see text for more details . ] in fig .
[ raa ] we show the rhic data for the @xmath38 , for various produced mesons @xcite and direct photons @xcite in central au+au collisions at midrapidity . a large suppression in high @xmath39 meson production is observed , and those for @xmath40 s being almost flat at @xmath41 up to 20 gev/@xmath42 .
the figure also shows that the level of suppression for @xmath40 s , @xmath43 s and @xmath18-mesons are very similar , which supports the conclusion that the suppression occurs in the partonic phase , not in the hadronic phase .
this strong suppression of meson production is in contrast to the behavior of direct photons , also shown in the figure .
the direct photons follow binary scaling ( i.e. @xmath44 ) or no suppression .
this is a strong evidence that the suppression is not an initial state effect , but a final state effect caused by the high density medium with color charges created in the collision .
this is further consolidated by a demonstration through a controlled experiment using deuteron on au ion collisions , which gave a @xmath45 @xmath3 1 for @xmath46 at midrapidity and high @xmath39 @xcite .
the various curves in the fig .
[ raa ] represents different model calculations .
the dashed curve shows a theoretical prediction using the glv parton energy loss model @xcite .
the model assumes an initial parton density @xmath47 , which corresponds to an energy density of approximately 5 - 15 gev/@xmath48 .
the lower dashed curves are for higher gluon density .
the precision high @xmath39 data at rhic has been used to characterize medium density fairly accurately .
the conclusion being that the medium formed in central au+au collisions at rhic has a high degree of opacity @xcite .
in addition , theoretical studies suggest that for a given initial density , the @xmath32(@xmath39 ) values are also sensitive to the lifetime ( @xmath49 ) of dense matter formed in heavy - ion collisions @xcite .
the solid curves are predictions from ref .
@xcite at @xmath36 = 200 gev with @xmath49 = 10 fm/@xmath42 ( i.e. larger than the typical system size of @xmath3 6 - 7 fm ) .
the parton energy loss calculations discussed above attributes the opacity to plasma induced radiation of gluons , much like ordinary bremsstrahlung of photons by electrons .
however , the quantitatively large suppression pattern observed at high @xmath39 , for both light hadrons and those involving heavy quarks @xcite , showed that the mechanism of energy loss is far from being a settled issue , namely , the relative contribution of radiative and collisional forms . as an example , shown in fig .
[ raa ] is a comparison of the data to theoretical results ( dot - dash curves ) on @xmath32 from models that consider only collisional energy loss @xcite .
this model gives @xmath32 values at high @xmath39 close to the measured values and similar to corresponding values from models having only a radiative mechanism for parton energy loss .
leading particle measurements , such as the ones shown in fig .
[ raa ] suffer from a number of limitations .
( a ) leading hadrons come from a mixture of parent quarks and gluons .
( b ) as a fragmentation product , the energy of a leading hadron is not a perfect proxy for the energy of the parent parton as it samples a wide range of partonic energies .
in future we should look forward to potentially three interesting measurements .
( i ) the @xmath50-jet process potentially provides access to the underlying scattered parton s energy .
measurements of the distribution of particles from the jet opposite , in azimuth , to the tagged photon reveals how much energy was lost , and how it was redistributed , by the colored parton as it traversed the medium @xcite .
( ii ) another method is through the full reconstruction of jets in heavy - ion collisions @xcite . beyond producing a far better proxy for the energy of the parent parton than a leading hadron
, this technique allows one to trace the evolution of energy flow in directions both longitudinal and transverse to the direction of the parent parton .
both these methods are under active pursual at rhic .
( iii ) another important feature of jet quenching is provided by partonic identity . while it is difficult to disentangle light quarks from gluons , especially in a heavy - ion environment , charm and bottom can be easily tagged by the existence of a charmed or bottom hadron in the final state . due to their large masses
the charm and bottom quarks are predominantly produced via hard scattering in the initial stage of the high - energy heavy - ion collision .
the final state spectra can therefore serve as a sensitive tool to probe in - medium rescattering and interactions responsible for thermalization .
this will also allow to study the non - abelian feature of qcd that results in the gluons losing more energy than quarks in the medium @xcite .
plans are in place to measure the cross - sections and transverse momentum spectra of hadrons with open and hidden heavy flavor at rhic with new detector upgrades .
this will also provide useful data to understand the different mechanisms of energy loss : collisional versus radiative .
elliptic flow , @xmath51 , is an observable which is thought to reflect the conditions from the early stage of the collisions @xcite . in non - central heavy - ion collisions ,
the initial spatial anisotropy of the overlap region of the colliding nuclei is transformed into an anisotropy in momentum space through interactions between the particles .
as the system expands it becomes more spherical , thus the driving force quenches itself .
therefore the elliptic flow is sensitive to the collision dynamics in the early stages .
it is measured , by calculating @xmath52 , where @xmath18 is the azimuthal angle of the produced particles and @xmath14 is the azimuthal angle of the impact parameter , and angular brackets denote an average over many particles and events . for ( a ) light quark hadrons and ( b ) strange quark hadrons .
the data is from the minimum bias au+au collisions at midrapidity for @xmath12 = 200 gev @xcite .
top right plot : compilation of the number of constituent quark scaled @xmath53 as a function of the scaled transverse kinetic energy @xcite .
bottom plot : compilation of @xmath54 extracted from various measurements in heavy - ion collisions at rhic .
, title="fig : " ] for ( a ) light quark hadrons and ( b ) strange quark hadrons .
the data is from the minimum bias au+au collisions at midrapidity for @xmath12 = 200 gev @xcite .
top right plot : compilation of the number of constituent quark scaled @xmath53 as a function of the scaled transverse kinetic energy @xcite .
bottom plot : compilation of @xmath54 extracted from various measurements in heavy - ion collisions at rhic .
, title="fig : " ] for ( a ) light quark hadrons and ( b ) strange quark hadrons .
the data is from the minimum bias au+au collisions at midrapidity for @xmath12 = 200 gev @xcite .
top right plot : compilation of the number of constituent quark scaled @xmath53 as a function of the scaled transverse kinetic energy @xcite .
bottom plot : compilation of @xmath54 extracted from various measurements in heavy - ion collisions at rhic .
, title="fig : " ] [ v2 ] figure 4 top left panel shows the rhic results on @xmath53 of light quark ( @xmath55 and @xmath15 ) and strange quark ( @xmath18 and @xmath24 ) carrying hadrons in au+au collisions at 200 gev @xcite .
three very distinct experimental observations can be made .
( a ) at low @xmath39 ( @xmath56 2 gev/@xmath42 ) the heavier hadrons have smaller @xmath51 .
such a mass ordering is expected in hydrodynamics calculations of @xmath51(@xmath39 ) for identified particles @xcite .
( b ) at the intermediate @xmath39 range of 2 - 5 gev/@xmath42 it is observed that baryons have higher @xmath51 than mesons .
the @xmath18-meson @xmath51 plays a crucial role in establishing this baryon - meson difference .
such a separation of baryons and mesons in the intermediate @xmath31 range has been also observed in measurements of the nuclear modification factor , @xmath57 @xcite .
these results are consistent with calculations from quark recombination models @xcite implying the de - confinement of the system prior to hadronisation .
( c ) comparison between the @xmath51 results for light quark carrying hadrons to those from strange quark carrying hadrons indicates both types of hadrons show similar magnitude of @xmath51 .
the multi - strange hadrons ( @xmath18 and @xmath24 ) have relatively low hadronic interaction cross sections and freeze - out early , hence are considered as the most promising probes of the early stages of the collision .
all the above results indicates that a substantial amount of collectivity has been developed at the partonic stage of the heavy - ion collisions .
in fact none of the available hadronic models are able to account for the observed magnitude of @xmath51 at @xmath36 = 200 gev .
only models which have additional partonic interactions explain the @xmath58 values @xcite . when elliptic flow @xmath53 is plotted versus transverse kinetic energy ( @xmath59 ) , both divided by the number of constituent quarks , the @xmath53 for all identified hadrons as well as light nuclei below ( @xmath59 ) @xmath3 1 gev/@xmath60 falls on a universal curve @xcite .
@xmath61 is the mass of the particle .
this scaling behavior as shown in fig.4 further strengthens the evidence for formation of partonic matter during the au + au collision process at 200 gev .
it is very hard to explain this observed pattern in a scenario where only hadronic matter exists throughout the interaction , whereas the hypothesis of coalescence of hadrons from de - confined quarks offers a ready explanation .
turn - off of the scaling at a given beam energy would indicate the hadronic side of the phase boundary .
the @xmath53 measurements of light quark carrying hadrons , the nuclear modification factor and @xmath53 for heavy quark carrying hadrons and the differential @xmath39 correlations for charged hadrons have been used to extract information of a dimensionless ratio , shear viscosity to entropy ( @xmath54 ) , for the medium formed in heavy - ion collisions at rhic @xcite .
figure 4 bottom panel shows the compilation of @xmath54 extracted from various measurements @xcite .
it is observed to lie between that conjectured from quantum theory ( @xmath54 @xmath3 1/4@xmath55 ) and those for liquid helium at @xmath62 .
such a low value of @xmath54 ( within a factor 1 - 10 of the quantum limit ) indicates that the matter formed in heavy - ion collisions at rhic has low viscosity , hence is a strongly coupled system .
the cp is a landmark point in the qcd phase diagram , observation of which will make the qcd phase diagram a reality .
a close collaboration between the experiments and theory perhaps will lead to its discovery .
the first step in this process is to establish an observable for cp which can be measured experimentally and can be related to qcd calculations . in this context , it is important to recall that for a static , infinite medium , the correlation length ( @xmath63 ) diverges at the cp .
@xmath63 is related to various moments of the distributions of conserved quantities such as net - baryons , net - charge , and net - strangeness @xcite .
typically variances ( @xmath64 @xmath65 @xmath66 ; @xmath67 ; @xmath68 is the mean ) of these distributions are related to @xmath63 as @xmath64 @xmath3 @xmath69 @xcite . finite size and time effects in heavy - ion collisions put constraints on the values of @xmath63 .
a theoretical calculation suggests @xmath63 @xmath70 2 - 3 fm for heavy - ion collisions @xcite .
it was recently shown that higher moments of distributions of conserved quantities , measuring deviations from a gaussian , have a sensitivity to cp fluctuations that is better than that of @xmath64 , due to a stronger dependence on @xmath63 @xcite .
the numerators in skewness ( @xmath71 = @xmath72 ) goes as @xmath73 and kurtosis ( @xmath74 = [ @xmath75 - 3 ) goes as @xmath76 . a crossing of the phase boundary can manifest itself by a change of sign of @xmath77 as a function of energy density @xcite .
dependence of @xmath78 , @xmath74@xmath64 and @xmath79 for net - proton distributions measured at rhic @xcite .
the results are compared lattice qcd calculations @xcite and hrg model @xcite . also shown on the top
are the temperature and baryon chemical potential values at chemical freeze - out extracted from particle yields using a thermal model . ]
the first connections between qcd calculations and experiment has been recently made @xcite .
lattice calculations and qcd - based models show that moments of net - baryon distributions are related to baryon number ( @xmath80 ) susceptibilities ( @xmath81 ; _ v _ is the volume ) @xcite .
then one can construct ratios such as : @xmath82 which do not contain the volume and therefore provide a direct and convenient comparison of experiment and theory . in the above expressions
the left hand side of each equality can be measured in an experiment whereas the right hand side can be calculated by lattice qcd .
close to the cp models predict the @xmath80 distributions to be non - gaussian and susceptibilities to diverge causing the experimental observables to have large values .
the experimental values should also be compared to those expected from statistics , for example if the @xmath15 and @xmath83 distributions are individually poissonian then @xmath74@xmath64 for net - protons is unity . experimentally measuring
event - by - event net - baryon number is difficult .
however , the net - proton multiplicity ( @xmath84 = @xmath85 ) distribution is measurable .
theoretical calculations have shown that @xmath85 fluctuations reflect the singularity of the charge and baryon number susceptibility as expected at the cp @xcite .
non - cp model calculations show that the inclusion of other baryons does not add to the sensitivity of the observable @xcite .
figure [ qcp ] shows the energy dependence of @xmath78 , @xmath74@xmath64 and @xmath79 for @xmath85 , compared to lattice qcd @xcite and hadron resonance gas ( hrg ) model which does not include a cp @xcite .
the experimental values plotted are for central au+au collisions for @xmath36 = 19.6 , 62.4 and 200 gev
. the lattice calculations , which predict a cp around @xmath11 @xmath3 300 mev , are carried out using two - flavor qcd with number of lattice sites in imaginary time to be 6 and mass of pion around 230 mev @xcite .
the ratios of the non - linear susceptibilities at finite @xmath11 are obtained using pad approximant resummations of the quark number susceptibility series .
the freeze - out parameters as a function of @xmath36 are from @xcite and @xmath86 = 175 mev . from comparisons of the experimental data to the hrg model and
the lack of non - monotonic dependence of @xmath74@xmath64 on @xmath36 studied , one concludes that there is no indication from the current measurements at rhic for a cp in the region of the phase plane with @xmath11 @xmath56 200 mev . although it must be noted that the errors on the experimental data point at 19.6 gev is quite large due to small event statistics .
it is difficult to rule out the existence of cp for the entire @xmath11 region below 200 mev .
the extent to which these results can do that is guided by the theoretical work . in addition , the expectation of the extent of the critical region in @xmath11 is thought to be about 100 mev . the results discussed here form the baseline for the future cp search program at rhic @xcite .
however the fact that the data shows excellent agreement with hrg and lattice qcd , both of which assume thermalization , is another non - trivial indication of attainment of thermalization ( some other measurements are discussed in next section ) in heavy - ion collisions .
such a conclusion is drawn for the first time using fluctuation measurements .
with the idea that the rise and then fall of observables sensitive to cp as @xmath10 increases should allow us to ascertain the ( @xmath0,@xmath10 ) coordinates of the critical point , the beam energy scan program at rhic has started .
the first phase of the experimental program at rhic is expected to be completed in 2010 - 2011 .
this phase is expected to cover a @xmath36 region of 39 to 5.5 gev , which corresponds to a @xmath10 range of 112 to 550 mev @xcite .
in central au+au collisions at @xmath36 = 200 gev @xcite .
the lines show the possibility of having three regions in the spectra where the dominant mechanism of particle productions are different .
see text for more details . ]
the measured hadron spectra reflect the properties of the bulk matter at kinetic freeze - out , after elastic collisions among the hadrons have ceased .
more direct information on the earlier stages can be deduced from the integrated yields of the different hadron species , which change only via inelastic collisions .
the point in time at which these inelastic collisions cease is referred to as chemical freeze - out , which takes place before kinetic freeze - out .
rhic experiments have measured the @xmath39 distribution of variety of particles over a wide range of @xmath39 at midrapidity . a sample of the invariant yield ( @xmath87 @xmath88 ) of produced particles at rhic for central au+au collisions at @xmath36 = 200 gev @xcite is shown in fig . [ spectra ] .
like the @xmath39 dependence of @xmath51 discussed in the section 2.3 , here also we can separate the spectra into 3 regions based on the dominant mechanism of particle production . the low @xmath39 ( @xmath56 2 gev/@xmath42 )
is explained by thermal model based calculations @xcite , intermediate @xmath39 ( 2 - 6 gev/@xmath42 ) by parton recombination based approaches @xcite and high @xmath39 ( @xmath28 6 gev/@xmath42 ) by including pqcd based processes or jet production @xcite .
the only statistical distribution which so far seems to sucessfully describe the @xmath31 spectra and @xmath53(@xmath89 ) over a wide momentum range is the one based on tsallis statistics @xcite . in this section
we concentrate on low @xmath39 part of the spectra for rest of the discussions .
the transverse momentum distributions of the different particles contain two components , one random and one collective .
the random component can be identified as the one that depends on the temperature of the system at kinetic freeze - out ( @xmath90 ) .
the collective component , which arises from the matter density gradient from the center to the boundary of the fireball created in high - energy nuclear collisions , is generated by collective flow in the transverse direction , and is characterized by its velocity @xmath91 .
assuming that the system attains thermal equilibrium , the blast - wave ( bw ) formulation @xcite can be used to extract @xmath90 and @xmath92 .
the transverse flow velocity of a particle at a distance @xmath93 from the center of the emission source , as a function of the surface velocity ( @xmath94 ) of the expanding cylinder , is parameterized as @xmath95 , where @xmath96 is found by fitting the data .
the transverse momentum spectrum is then @xmath97 where @xmath98 and @xmath99 are modified bessel functions and @xmath100 .
contours , extracted from thermal + radial flow fits ( without allowance for resonance feed - down ) , for produced hadrons @xmath101 and @xmath15 and multi - strange hadrons @xmath18 and @xmath24 . on the top of the plot , the numerical labels indicate the centrality selection . for @xmath101 and @xmath15 , 9 centrality bins ( from top 5% to 70 - 80% ) were used for @xmath102 gev au+au collisions @xcite .
the results from p+p collisions are also shown . for @xmath18 and @xmath24 ,
only the most central results are presented . dashed and
solid lines are the 1-@xmath103 and 2-@xmath103 contours , respectively .
right plot : ratios of @xmath104-integrated mid - rapidity yields for different hadron species measured in star for central au+au collisions at @xmath30 = 200 gev .
the horizontal bars represent statistical model fits to the measured yield ratios for stable and long - lived hadrons .
the fit parameters are @xmath105 mev , @xmath106 mev , @xmath107 .
the variation of @xmath108 with centrality is shown in the inset , including the value ( leftmost point ) from fits to yield ratios measured by star for 200 gev p+p collisions .
, title="fig : " ] contours , extracted from thermal + radial flow fits ( without allowance for resonance feed - down ) , for produced hadrons @xmath101 and @xmath15 and multi - strange hadrons @xmath18 and @xmath24 . on the top of the plot , the numerical labels indicate the centrality selection . for @xmath101 and @xmath15 , 9 centrality bins ( from top 5% to 70 - 80% ) were used for @xmath102 gev au+au collisions @xcite .
the results from p+p collisions are also shown . for @xmath18 and @xmath24 ,
only the most central results are presented .
dashed and solid lines are the 1-@xmath103 and 2-@xmath103 contours , respectively .
right plot : ratios of @xmath104-integrated mid - rapidity yields for different hadron species measured in star for central au+au collisions at @xmath30 = 200 gev .
the horizontal bars represent statistical model fits to the measured yield ratios for stable and long - lived hadrons .
the fit parameters are @xmath105 mev , @xmath106 mev , @xmath107 .
the variation of @xmath108 with centrality is shown in the inset , including the value ( leftmost point ) from fits to yield ratios measured by star for 200 gev p+p collisions .
, title="fig : " ] fits to the identified hadrons @xmath31 distributions at midrapidity for au+au collisions at @xmath30 = 200 gev using eq .
[ blasteq ] are carried out .
the extracted model parameters characterizing the random ( generally interpreted as a kinetic freeze - out temperature @xmath109 ) and collective ( radial flow velocity @xmath92 ) are shown in fig .
[ kfo ] in terms of confidence level ( @xmath110 ) contours , for various impact parameters of the collision .
as the collisions become more and more central , the bulk of the system , dominated by the yields of @xmath111 have lower kinetic freeze - out temperature and develops stronger collective flow . on the other hand , even for the most central collisions , the spectra for multi - strange particles @xmath18 and @xmath24 appear to reflect a higher freeze - out temperature . within a statistical model in thermodynamical equilibrium ,
the particle abundance in a system of volume @xmath112 can be given by @xmath113 where @xmath114 is the abundance of particle species @xmath115 , @xmath116 is the spin degeneracy , @xmath117 and @xmath118 are the baryon number and strangeness number , respectively , @xmath119 is the particle energy , and the integral is taken over all momentum space @xcite .
the model parameters are the chemical freeze - out temperature ( @xmath120 ) , the baryon ( @xmath121 ) and strangeness ( @xmath122 ) chemical potentials , and the _ ad hoc _ strangeness suppression factor ( @xmath123 ) .
measured particle ratios are used to constrain the values of @xmath120 and @xmath11 at chemical freeze - out .
figure [ kfo ] compares star measurements of integrated hadron yield ratios for central au+au collisions to statistical model calculations .
the excellent agreement between data and model are observed .
the ratios which include stable and long - lived hadrons through multi - strange baryons , is consistent with the light flavors , @xmath124 and @xmath16 , having reached chemical equilibrium ( for central and near - central collisions only ) at @xmath125 mev .
the deviations of the short - lived resonance yields , such as those for @xmath126 and k@xmath127 collected near the right side of fig .
[ kfo ] , from the statistical model fits , presumably result from hadronic re - scattering after the chemical freeze - out and needs to be further understood .
the saturation of the strange sector yields , attained for the first time in near - central rhic collisions , is particularly significant .
the saturation is indicated quantitatively by the value obtained for the non - equilibrium parameter @xmath108 for the strange sector for central collisions .
the temperature deduced from the fits is essentially equal to the critical value for a qgp - to - hadron - gas transition predicted by lattice qcd @xcite , but is also close to the hagedorn limit for a hadron resonance gas , predicted without any consideration of quark and gluon degrees of freedom @xcite .
if thermalization is indeed achieved by the bulk matter prior to chemical freeze - out , then the deduced value of @xmath128 represents a lower limit on that thermalization temperature .
) from heavy - ion collisions at various @xmath30 @xcite .
the @xmath121 values shown are estimated at chemical freeze - out .
the kinetic and chemical freeze - out parameters , extracted using models assuming thermal and chemical equilibrium from midrapidity particle ratio and @xmath89 spectra measurements in heavy - ion collisions . the range of critical temperatures ( @xmath129 ) of the cross - over quark - hadron phase transition at @xmath121 = 0 and the qcd critical point from two different calculations from lattice qcd are also indicated .
model - based estimates of the range of initial temperature ( @xmath130 ) achieved in heavy - ion collisions based in part on direct photon data at top rhic @xcite and sps @xcite energies are also shown .
the range of @xmath121 to be scanned in the rhic beam energy scan program corresponding to @xmath30 = 5.5 to 39 gev as well as experiments at sps and cbm are indicated as shaded ellipse .
the solid point around @xmath0 @xmath3 0 and @xmath11 = 938 mev represents nuclear matter in the ground state . ] in summary , the current understanding of the qcd phase diagram is depicted in fig .
[ phase ] . from the qcd calculations on lattice
it is now established theoretically that the quark - hadron transition at @xmath1 = 0 is a cross - over .
the critical temperature for a quark - hadron phase transition lies within a range of 170 - 190 mev .
most calculations on lattice also indicate the existence of qcd critical point for @xmath1 @xmath28 160 mev .
the exact location of cp is not yet known unambiguously .
two such predictions computed on lattice are shown in fig .
[ phase ] for a @xmath62 of 176 mev @xcite .
new distinct signatures have been predicted by qcd model calculations to locate the critical point in the phase diagram .
the specific suggestion for a cp search is to look for non - monotonic variation in the products of the higher moments of net - proton and net - charge distributions , which are related to susceptibilities , as a function of @xmath30 ( or @xmath0,@xmath1 ) . at top rhic energies , fluctuations in net - proton numbers
are found to be consistent with expectations from hrg and lattice qcd calculations @xcite .
these measurements established the baseline for the cp search in the heavy - ion collision program , put constraints on the location of cp in the qcd phase plane and added to the evidences for thermalization in central au+au collisions at rhic .
high energy heavy - ion collision experiments have seen distinct signatures which suggest that the relevant degrees of freedom in the initial stages of the collisions at top rhic energies are quark and gluons @xcite .
three such signatures related to strangeness enhancement , jet quenching and partonic collectivity are discussed in this paper .
the initial temperatures ( @xmath130 ) achieved at top rhic and sps energies are obtained from models @xcite that explain the direct photon measurements from the phenix experiment at rhic @xcite and from the wa98 experiment at sps @xcite . from these models , which assume that thermalization is achieved in the collisions within a time between 0.11.2 fm/@xmath42
, the @xmath130 extracted is greater than 300 mev at rhic and greater than 200 mev at sps .
further , the understanding of suppression in high @xmath31 hadron production in heavy - ion collisions relative to @xmath15+@xmath15 collisions at rhic requires a medium energy density @xmath131 1 gev/@xmath48 ( critical energy density from lattice for a phase transition ) .
this also shows that the medium has a high degree of opacity to propagation of color charges .
in addition the measurement of elliptic flow and the observation of number of constituent quark scaling demonstrates that substantial collectivity has been developed in the partonic phase .
the magnitude of the flow across several hadronic species and a small value of viscosity to entropy ratio extracted from the data supports the idea of formation of a strongly coupled system in the heavy - ion collisions .
this then also supports the notion of creating a liquid with low viscosity in high energy nuclear collisions @xcite .
the experiments have also measured the temperature at which the inelastic collisions ceases ( chemical freeze - out ) and elastic collisions ceases ( kinetic freeze - out ) . these temperatures ( as shown in fig .
[ phase ] ) are extracted from the measured particle ratios and transverse momentum distributions using model calculations which assume the system is in chemical and thermal equilibrium .
new experimental programs at rhic , sps , fair and nica facilities have been designed to explore a large part of the qcd phase digram , covering a @xmath11 range of 20 - 600 mev . whereas the experimental program at lhc ( probing the cross over region of @xmath1 @xmath3 0 mev of the phase diagram ) have started to provide an unique opportunity to understand the properties of matter governed by quark - gluon degrees of freedom at unprecedented high initial temperatures ( higher plasma life time ) achieved in the pb+pb collisions at 2.76 - 5.5 tev @xcite . both at lhc and rhic , one specific observable that has the potential to provide a further understanding of system formed in heavy - ion collision are the dileptons .
theoretically they are from the virtual photons , and are different from real photons in having a mass .
the dilepton mass opens up a new dimension and can be used to study time evolution of the system in heavy - ion collisions .
for example , recently it has been discussed that virtual photon ( dilepton ) interferometry provide access to the development of collective flow with time @xcite .
studying the @xmath89 dependence of the elliptic flow and nuclear modification factor for dileptons for masses corresponding to various hadrons and beyond will help us understand partonic collectivity and medium opacity .
comparing the spectral functions of resonances decaying to dileptons and hadrons will let us know about the medium effects .
while the slope of the dilepton @xmath89 distributions will tell us about development of radial flow and provide direct evidence of thermal radiation of partonic origin in high energy nuclear collisions @xcite .
i would like to thank drs .
j. alam , s. chatopadhyay , s. gupta , v. koch , l. kumar , t. k. nayak , p. k. netrakanti , d. p. mahapatra , k. rajagopal , h. g. ritter , m. stephanov , y. p.
viyogi , n. xu and z. xu for useful discussions .
i would like to thank c. jena , x. f. luo , md .
nasim , s. singha and drs .
m. sharma , s. shi and a. tang for their help in preparation of this manuscript .
this work is supported by dae - brns project scantion no .
10 brahms collaboration , i. arsene et al . , nucl . phys . a 757 ( 2005 ) 1
; phobos collaboration , b.b . back
et al . , nucl .
phys . a 757 ( 2005 ) 28 ; star collaboration , j. adams et al . , nucl .
phys . a 757 ( 2005 ) 102
; phenix collaboration , k. adcox et al . ,
phys . a 757 ( 2005 ) 184 .
m. stephanov , prog .
153 , 139 ( 2004 ) ; int .
a20 , 4387 ( 2005 ) ; z. fodor and s.d .
katz , jhep0404 , 50 ( 2004 ) ; r. v. gavai and s. gupta , phys . rev .
d78 , 114503(2008 ) ; phys .
d71 , 114014 ( 2005 ) .
phenix collaboration , s. s. adler et al .
91 , 072301 ( 2003 ) ; phobos collaboration , b. b. back et al .
91 , 072302 ( 2003 ) ; star collaboration , j. adams et al.,phys .
91 , 072304 ( 2003 ) ; brahms collaboration , i. arsene et al .
, phys . rev .
91 , 072305 ( 2003 ) .
star collaboration , b. i. abelev et al .
97 , 152301 ( 2006 ) ; phys .
b 655 , 104 ( 2007 ) ; phenix collaboration , a. adare et al . , arxiv:1004.3532 ; phys . rev .
c 82 , 011902(r ) ( 2010);phys
101 , 232301 ( 2008 ) .
n. armesto et al . , phys . rev .
d 71 , 054027 ( 2005);x .- n .
wang , phys .
c 58 , 2321 ( 1998 ) ; q. wang and x .-
wang , phys .
c 71 , 014903 ( 2005 ) ; b. mohanty ( for star collaboration ) , arxiv:0705.0953 .
d. teaney , j. lauret , e. v. shuryak , phys .
86 , 4783 ( 2001 ) .
p. huovinen , p.f .
kolb , u. heinz , p.v .
ruuskanen , and s.a .
voloshin , phys .
b 503 , 58 ( 2001 ) ; c. nonaka et al . , phys
b 583 , 73 ( 2004 ) .
t. hirano and y. nara , phys .
c 69 , 034908 ( 2004 ) .
d. molnar and s.a .
voloshin , phys .
91 , 092301 ( 2003 ) ; r. j. fries et al . ,
c 68 , 044902 ( 2003 ) ; j.h . chen et al . ,
c 74 , 064902 ( 2006 ) ; v. greco , c.m .
ko , p. levai , phys .
90 , 202302 ( 2003 ) .
roy a. lacey et al .
98 , 092301 ( 2007 ) ; s. gavin and m. abdel - aziz , phys .
97 , 162302 ( 2006 ) ; h - j .
drescher , a. dumitru , c. gombeaud , j .- y .
ollitrault , phys .
c 76 , 024905 ( 2007 ) .
r. chatterjee , d. k. srivastava and s. jeon , phys .
c 79 , 034906 ( 2009 ) ; p. huovinen , p.v .
ruuskanen and s. s. rasanen , phys .
b 535 , 109 ( 2002 ) ; a. k. chaudhuri , j. phys .
g 29 , 235 ( 2003 ) ; j. alam et al .
c 63 , 021901 ( 2001 ) ; d. denterria and d. peressounko , eur .
j. c 46 , 451 ( 2006 ) ; j. alam et al . , j. phys .
g 34 , 871 ( 2007 ) .
|
quantum chromodynamics ( qcd ) phase diagram is usually plotted as temperature ( @xmath0 ) versus the chemical potential associated with the conserved baryon number ( @xmath1 ) .
two fundamental properties of qcd , related to confinement and chiral symmetry , allows for two corresponding phase transitions when @xmath0 and @xmath1 are varied .
theoretically the phase diagram is explored through non - perturbative qcd calculations on lattice .
the energy scale for the phase diagram ( @xmath2 @xmath3 200 mev ) is such that it can be explored experimentally by colliding nuclei at varying beam energies in the laboratory . in this paper
we review some aspects of the qcd phase structure as explored through the experimental studies using high energy nuclear collisions . specifically , we discuss three observations related to the formation of a strongly coupled plasma of quarks and gluons in the collisions , experimental search for the qcd critical point on the phase diagram and freeze - out properties of the hadronic phase
. _ keywords _ : quark gluon plasma , strangeness enhancement , jet quenching , elliptic flow , chemical and kinetic freeze - out , qcd phase digram and qcd critical point .
|
since the first experimental observation of bose - einstein ( be ) condensation@xcite in bose atom clouds in a harmonic trap , both experimental and theoretical studies of this phenomenon and various properties of the condensate has grown up to a fast expanding field of research @xcite .
while studies of bose condensation in a harmonic potential is of great interest , investigations of the be condensation of lattice - bosons and free - bosons in _ anharmonic _ potentials , in particular in a quartic potential , have received considerable attention in recent times@xcite ( see also the note in ref .
@xcite ) . all the previous works@xcite , except two@xcite , deals with _ lattice - bosons _ in anharmonic confinements .
the previous works on free - bosons in power law traps and quartic traps in three and lower dimensions@xcite dealt with the calculations of bose condensation temperature for _ non - interacting _ bosons in those environments . in this paper , we consider _ interacting _ bosons in a quartic potential .
the study of _ interacting _ bosons in a quartic potential may have important implications for fields ranging from chaotic dynamics to high energy physics@xcite .
further , it would be desirable to have minimum inhomogeneity in the density profiles for clear observation of several phases and transitions possible for interacting boson , fermion , and boson - fermion models@xcite .
while interactions in a many - boson system are expected to broaden the real space distribution of bosons in the confining region , it is not clear how far , from the center , a homogeneous distribution of bosons will exist .
we will show , among other results , that the interaction leads to a more or less homogeneous boson distribution in a substantial region ( compared to the non interacting bosons case ) in and around the trap center .
it would also be of contemporary interest to investigate the dynamics of the condensate in the specific quartic potential environment and compare it with the usually employed quadratic environment .
it may be expected that the experiments would first involve investigations of interacting bosons in a quartic potential before an additional lattice potential is applied .
furthermore , experimental studies of interacting bosons in the simplest anharmonic environment may be of considerable interest in its own right . within this general setting , it would be useful to make a detailed theoretical study of _ interacting _ bosons in a quartic potential .
it is the purpose of this paper to present some results of such an investigation using a combination of analytical and numerical methods .
this paper is organized in the following three sections . in sec
[ sec2 ] , the condensate properties are analyzed through a study of the gross - pitaevskii equation@xcite employing the thomas - fermi approximation@xcite . in sec
[ sec3 ] we present a comparative study of real space distributions of bosons in condensates of interacting bosons in quartic and harmonic confinements , by numerically solving the gross - pitaevskii equation .
the time evolution of some properties ( various components of energy and width oscillations of the condensate ) of the condensate is given in sec [ sec4 ] , and the conclusions are presented in sec [ sec5 ] .
the ground state properties of a bose - einstein condensate ( bec ) is well described by a macroscopic wave - function @xmath0 , the time evolution of which is governed by the gross - pitaeveskii equation @xcite ( gpe ) . incorporating the trap potential as well as a repulsive interaction between the atoms forming the condensate ,
the gpe in three dimensions ( 3d ) is given by @xmath1 where @xmath2 , @xmath3 is the atomic mass , @xmath4 is the quartic trapping potential , @xmath5 is the total number of atoms forming the condensate , @xmath6 is the strength of interaction between atoms in the condensed state , and @xmath7 ( @xmath8 ) is the s - wave scattering length . before applying the thomas - fermi approximation , we non - dimensionalize eq .
( [ nls_eqn ] ) through a set of linear transformations : @xmath9 . after dropping the wiggles on the symbols ,
we obtain @xmath10 where @xmath11 in the absence of the non - linear interaction term , the gpe is reduced to a schrdinger equation ( with a quartic potential ) whose approximate eigenvalues can be obtained through perturbation theory or wkb approximation @xcite .
expressing @xmath12 in units of @xmath13 , we perform dimensional reductions @xcite to obtain corresponding gpes in lower dimensions . in general , the gpes for d = 1 , 2 , and 3 can be summarized in a single equation as @xmath14 where @xmath15 here @xmath16 and @xmath17 are scaling factors introduced to write gpe in 2d and 1d , respectively . in order to find a stationary solution of eq .
( [ nls_non_dim_eqn ] ) , we do a separation of variables @xmath18 $ ] , where @xmath19 is the chemical potential . starting from eq .
( [ nls_non_dim_eqn ] ) , we obtain @xmath20 in the preceding equation @xmath21 . for simplicity
we have neglected the cross terms from the potential ( see , however , sec .
iii - c where these terms are shown to have only minor effects on the properties we study ) . to analytically solve for the ground state of the stationary state gpe , as given in eq .
( [ stationary_state_eqn ] ) , we make use of the thomas - fermi approximation @xcite . for sufficiently large number of atoms ,
the kinetic energy is very small compared to the potential and the interaction energies .
so , a good approximation to the ground state is obtained by solving the gpe without the kinetic energy term . on using this approximation
, the gpe becomes @xmath22 the relevant solutions of the preceding cubic equation are @xmath23 since @xmath24 vanishes for @xmath25 , the thomas - fermi radius @xmath26 is obtained as @xmath27 in which the chemical potential @xmath28 is be determined from the normalization condition @xmath29 where @xmath30 = @xmath31 , @xmath32 , and @xmath33 for @xmath34 = 1 , 2 , and 3 , respectively .
then , the chemical potential is found to be @xmath35^\frac{4}{d+4}.\ ] ] so , for the 3d case , the chemical potential varies as @xmath36 for the quartic trap . in comparison
, we note that it varies as @xmath37 for bosons in a harmonic trap@xcite ( for which @xmath38 ) .
it may be noted that the @xmath5 dependence of the chemical potential becomes more dominant as the dimensionality _
decreases_. the scaling factors ( @xmath16 and @xmath17 ) were introduced in 2d and 1d gpe to obtain gpes in 2d and 1d .
they also ensure that the size of the condensate remains the same in all the three dimensions . to find @xmath16 and @xmath17
, we calculate @xmath39 in 2d and 1d and equate it to the chemical potential obtained in 3d@xcite . using this procedure
, we obtain @xmath40 @xmath41 and @xmath42
in this section , we present results on the spatial distribution of bosons in the condensate in 1d by numerically solving the gpe .
we solve the gpe by using the finite difference crank - nicholson ( fdcn ) method@xcite . in this method for the gpe given by @xmath43 the solution
is advanced in small time steps @xmath44 as given below @xmath45 the ground state solution is found by the imaginary time propagation method@xcite . in the imaginary time propagation method
, we replace @xmath44 by @xmath46 and propagate a initial trial wave function using the above mentioned propagation scheme . after suitable normalization at each time
step the wave - function converges to the ground state .
we have taken the space step as @xmath47 and the time step as @xmath48 .
for illustrative purpose , we have used a set of parameters for @xmath49 : @xmath50 , @xmath51 , @xmath52 . in fig . 1
, we show the condensate density profiles for systems with @xmath53 to @xmath54 bosons in a quartic trap . further
, this figure also contains results of the tf approximation for the same numbers of bosons in a quartic trap .
furthermore , we have also shown the results for bosons in a harmonic potential .
we find that the tf approximation results are in quantitative agreement with the numerical solutions of gpe when the number of bosons is large . on increasing the number of atoms , the width of the ground state increases , which is similar to the case of bosons in a harmonic potential@xcite , except that the spread of condensate is larger for a harmonic potential .
we also notice that the ground state density profiles for the quartic trap becomes relatively flat with increasing number of bosons in the quartic trap .
next , we fix the number of atoms to @xmath55 in a quartic trap and investigate the boson distribution in the ground state for different interaction strengths .
the results are displayed in the left panel of fig .
we find that the ground state boson distribution is more spread out with increasing interaction strength .
it may be argued that the redistribution of bosons from the center to other parts of the condensate is due to the peak in the non - interacting distribution in the central region since the @xmath56 wave - function is peaked at the center and is symmetric around the center .
but , for the interacting bosons case , the interaction term is to be calculated in a self - consistent manner including the effect of the interaction .
hence , it is not obvious that interacting bosons distribution will have the profile shown in fig .
2 . it is the result of a combined effect of interaction and self - consistent changes in the wave - function .
thus , we see that the interacting bosons , compared to the non - interacting bosons , have a density profile which is more or less uniform in the central region in the case of a quartic confining potential . in fig .
2 , we have also exhibited the ground state chemical potential for different values of the interaction parameter @xmath57 .
the chemical potential for a condensate in a quartic trap is larger than that in harmonic trap , the difference becoming more significant for higher values of @xmath57 .
we have solved the gpe in higher dimensions by using the split - step fdcn method@xcite . in this method , in 3d , the hamiltonian @xmath58 is split into different non - derivative and derivative parts @xmath59 , @xmath60 , and @xmath61 , where @xmath62 @xmath63 the time evolution is performed in the following steps .
let @xmath64 be the wave - function at time @xmath65 .
this wave - function is advanced first over a time step @xmath44 at @xmath65 to produce an intermediate solution @xmath66 from @xmath64 via @xmath67 next the time propagation is performed via the following semi - implicit crank- nicholson scheme : @xmath68 @xmath69 and @xmath70 an analogous set of equations hold in 2d . in the calculations , we have used space steps of 0.1 ( for @xmath71 and @xmath72 in 2d and @xmath71 , @xmath72 , and @xmath73 in 3d ) and the time step @xmath48 . to calculate the ground state ,
again we use imaginary time propagation method by replacing @xmath44 with @xmath46 .
we start with an initial trial wave - function which is propagated via the above mentioned scheme . after suitable normalization , at each time step , the wave - function converges to the ground state . in fig .
3 , the ground state density profiles of the condensate are shown for different numbers of atoms ranging from @xmath74 to @xmath54 . on increasing the number of atoms , the width of the ground state increases , which is similar to that in 1d . on comparing the ground states for harmonic and quartic traps as shown in fig .
4 , we found that the ground state density profiles are flat in the central region for the quartic trap compared to the harmonic trap .
similar results are obtained for the 3d case as well . clearly , the quartic confinement is more advantageous to the quadratic one if one is interested in investigations of the properties of a bose condensate with a minimal influence of the confining potential .
we may note here that , in an earlier work@xcite , we have shown that the finite temperature properties of _ lattice _ bosons are closer to the pure lattice case if one uses quartic rather than a quadratic confining potential .
the effect of changing interaction strength on the ground state density profiles in a quartic trap are displayed in figs .
5 and 6 for 2d and 3d , respectively .
it is clear that the bosons gain energy by spreading into larger area in the traps with increasing strength of the interaction . in fig . 7 ,
the ground state chemical potential is plotted for different values of the interaction parameter @xmath75 for @xmath34 = 2 and 3 .
the chemical potentials of the condensate in a quartic traps are higher than those in harmonic traps , the difference becoming more significant at higher values of @xmath75 . in the preceding section
, we have presented results of our calculations neglecting the cross terms in the full quartic potential ( @xmath76 ) . in this section
, we investigate the effects of the previously neglected cross terms .
the contribution of the cross terms is in cross - coupling potentials across different dimensions .
one may expect that the inclusion of these cross - terms may lead to complex inhomogeneous density profiles .
but , on performing the calculations for the quartic trap , we find that the effects of cross - terms are rather minor .
they lead to small increases in the chemical potential and the number of bosons in the central region of the trap as shown in fig . 8 for 2d .
we do not find any significant inhomogeneities in the ground state condensate profiles generated by the cross - terms .
we have observed ( figures not shown ) similar effects in 3d as well .
it may be noted here that cross terms lead to a @xmath77 increase of the bose condensation temperature ( @xmath78 ) in 2d , and a @xmath79 increase in 3d@xcite .
in this section , we study the time evolution of the condensate after removal of the trapping potential and width oscillations of the condensate . to theoretically study the expansion of condensate , we time evolve the gpe after removing the potential term , taking the stationary ground state as the initial state function at @xmath80 . to time evolve the gpe we use the fdcn scheme given by eqs .
[ gpe cn]-[cn1d ] in 1d and split - step fdcn scheme as given in eqs .
[ non derivarive]-[final time ] in higher dimensions . here
the propagation is done in real time with time - step @xmath81 and space steps of @xmath82 ( for @xmath71 , @xmath72 , and @xmath73 ) .
the different energy components we calculate are given by@xcite : @xmath83 and , the the chemical potential is given by@xcite : @xmath84 the variation of various energy components and the chemical potential , for bosons in quartic and harmonic traps , with time is shown in fig . 9 for 1d and 2d .
similar results ( not shown ) are observed in 3d .
as shown in these figures , the chemical potential and the interaction energy decreases on time evolution , but there is an increase in kinetic energy term .
most of the energy _ before _ the expansion ( @xmath85 ) is contained in the mean - field and the potential energy of the bosons .
at @xmath80 , the confining potential is removed and the corresponding energy term from then on is zero . during the expansion , the chemical potential decreases due to decrease in the mean field .
further , the increase in kinetic energy during the expansion is due to transfer from the mean - field .
we have also shown in fig .
9 , the release energy ( @xmath86 ) which is a a time independent quantity .
the time evolution of the width of the condensate is displayed in fig . 10 for 1d , 2d , and 3d .
the condensate widths along three axes x , y , and z are defined as @xmath87 , @xmath88 , and @xmath89 , where brackets @xmath90 represent the expectation value at a particular instant during the time evolution .
we assume that at @xmath80 , the condensate is in a non - interacting ground state .
once the interaction is switched on at @xmath80 , it produces oscillations in the condensate . the fig .
10 show that the frequencies of oscillations are greater in the quartic potential case than in the harmonic case@xcite . for isotropic quartic potential , amplitude and frequency of oscillations in x , y , and z
directions are the same .
in this paper , we studied the static and dynamic properties of a be condensate in a quartic trap in one , two , and three dimensions .
the analysis was done using both analytical and complementary numerical techniques .
the analytical solutions were obtained using the thomas - fermi approximation .
the solutions obtained were then compared with the results obtained by numerically solving the gpe .
we compared the ground state density profiles of the condensate in a quartic trap for different strengths of interaction by changing the scattering length .
it was found that the thomas - fermi approximation results were in remarkably good agreement with the results obtained from the gpe for a large number of atoms in the condensate in a quartic confining potential .
the results were also compared with corresponding results for a bose condensate in a harmonic confining potential .
we find that the condensate density profiles are relatively flat in the central region of the quartic potential compared to the case of a condensate in a harmonic potential .
an implication of this result is that for investigating properties of a bose condensate with minimal influence of the overall confining potential , the quartic potential is more preferable compared to a quadratic potential .
we also find that increasing the number of atoms in the condensate or increasing the interactions spreads the condensate in the confining region .
our calculations of the chemical potential in the ground state show that it is larger for a condensate in a quartic trap compared to that in a harmonic trap .
the preceding conclusions were arrived at without including the cross - terms in a general quartic potential . on including the cross - terms
, we find that their effects lead to increases in the chemical potentials and condensate density in the central region of the trap .
these changes were found to be rather minor . from our results for the time evolution of the various energy components of an expanding condensate ( after removing the confining potentials ) , we find that the chemical potential and the potential energy decrease during the expansion while the kinetic energy increases .
it was also noticed that changes in the case of quartic potential are larger than the case of a harmonic potential .
the condensate width oscillations produced by switching on of the interaction , starting with a condensate of non - interacting bosons , were also studied .
it was found that the width oscillations of a condensate in a quartic trap have higher frequency than that in a harmonic trap .
gopesh kumar chaudhary thanks ugc , government of india , for financial support during this work .
99 m. h. anderson , j. r. ensher , m. r. matthews , c. e. weiman , and e. a. cornell , science * 269 * , 198 ( 1995 ) .
k. b. davis , m. -o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn , and w. ketterle , phys .
rev . lett . * 75 * , 3969 ( 1995 ) . c. c. bradley , c. a. sackett , j. j. tollett , and r. g. hulet , phys .
rev . lett . * 75 * , 1687 ( 1995 ) .
f. dalfovo , s. giorgini , l. p. pitaevskii , and s. stringari , rev .
phys . * 71 * , 463 ( 1999 ) .
a. j. leggett , rev .
phys . * 73 * , 307 ( 2001 ) .
j. r. anglin and w. ketterle , nature * 416 * , 211 ( 2002 ) . c. j. pethick and h. smith , _ bose - einstein condensation in dilute gases _
( cambridge university press , cambridge uk , 2002 ) . i. bloch , nature phys . * 1 * , 23 ( 2005 ) .
i. bloch , j. dalibard , and w. zwerger , rev .
phys . * 80 * , 885 ( 2008 ) .
v. i. yukalov , laser physics * 19 * , 1 ( 2009 ) .
o. gygi , h. g. katzgraber , m. troyer , s. wessel , and g. g. batrouni , phys .
a * 73 * , 063606 ( 2006 ) .
r. ramakumar and a. n. das , eur .
j. d * 47 * , 203 ( 2008 ) .
e. lundh , eur .
j. d * 46 * , 517 ( 2008 ) .
n. ma , k. y. yang , l. pollet , j. v. porto , m. troyer , and f. c. zhang , arxiv:0803.0546v1 [ cond-mat.other ] . v. bagnato and d. kleppner , phys . rev .
a * 44 * , 7439 ( 1991 ) .
s. gautam and d. angom , eur .
j. d * 46 * , 151 ( 2008 ) .
we note here that a large body of work has been done for bosons in mixed harmonic and quartic traps , starting with the suggestion of fetter that large vortices can be stabilized in fast rotating bose condensates in a harmonic trap by adding a quartic part to the confining potential [ see , for example , a. l. fetter , phys .
a * 64 * , 063608 ( 2001 ) ; v. bretin , s. stock , y. seurin , and j. dalibard , phys . rev . lett . *
92 * , 050403 ( 2004 ) , and references there in ] .
e. p. gross , nuovo cimento * 20 * , 454 ( 1961 ) .
l. p. pitaevskii , zh .
. fiz . * 40 * , 646 ( 1961 ) [ sov .
jetp * 13 * , 451 ( 1961 ) ] .
l. h. thomas , math .
cambridge philos .
* 23 * , 542 ( 1927 ) .
e. fermi , z. phys .
* 48 * , 73 ( 1928 ) .
p. m. mathews , m. seetharaman , and s. raghavan , j. phys .
a * 15 * , 103 ( 1982 ) .
j. e. williams , _ the preparation of topological modes in a strongly - coupled two - component bose - einstien condensate _ ( ph .
d. thesis , university of colorado , boulder , 1999 ) .
g. baym and c. j. pethick , phys .
lett . * 76 * , 6 ( 1996 ) .
s. k. adhikari and p. muruganandam , j. phys .
b : at . mol .
. phys . * 35 * , 2831 ( 2002 ) .
w. bao and w. tang , j. comput . phys . * 187 * , 230 ( 2003 ) .
p. muruganandam and s. k. adhikari , j. phys .
b : at . mol .
. phys . * 36 * , 2501 ( 2003 ) .
m. j. holland , d. s. jin , m. l. chiofalo , and j. cooper , phys .
lett . * 78 * , 3801 ( 1997 ) .
w. bao , d. jaksch , and a. markowich , j. comput . phys .
* 187 * , 318 ( 2003 ) .
|
in this paper , we present a theoretical study of a bose - einstein condensate of interacting bosons in a quartic trap in one , two , and three dimensions . using thomas - fermi approximation , suitably complemented by numerical solutions of the gross - pitaevskii equation
, we study the ground sate condensate density profiles , the chemical potential , the effects of cross - terms in the quartic potential , temporal evolution of various energy components of the condensate , and width oscillations of the condensate .
results obtained are compared with corresponding results for a bose condensate in a harmonic confinement .
|
the standard model of particle interactions is very successful , and the recent discovery of the higgs boson seals its validity . yet there are unanswered questions which on one side suggest to go `` beyond the standard model '' , perhaps connecting with a theory of quantum gravity , and on the other side beg for an `` explanation '' of the loose conceptual aspects , such as the hierarchy problem or the nature of symmetries . the spectral approach to noncommutative geometry @xcite provides a framework for the description of the standard model encoding it in a general view of geometry based on an algebraic description .
in particular , the action of a field theory is encoded in such a description , and one can construct a natural action for fermions and bosons based on the spectral properties of the ( generalized ) dirac operator .
such a _ spectral action _ has been introduced in ref .
@xcite and applied to the standard model in various forms ( for a recent review see ref .
this action is immediately applicable to the phenomenology and has been presently refined to confront itself with experimental results .
nevertheless , it is not free from conceptual issues . in this paper
we will mostly deal with the latter and try to solve some of these drawbacks with the introduction of a new form of action , the _
@xmath8 spectral action_. in the following , in order to distinguish the @xmath9 spectral action from the usual one , we will call the latter the _ cutoff spectral action _ since the main difference lies in the regularization procedure .
the key difference between the two actions lies in the fact that no operators of dimension higher than four appear in the @xmath8 spectral action , and therefore the theory is _
renormalizable_. in particular , it is not necessary to have to consider it as an effective theory valid just below the unification scale , and one can safely use it up to the planck scale where the very nature of spacetime changes due to quantum gravitational effects .
the ultraviolet asymptotics of the cutoff spectral action was discussed in ref .
@xcite , finding the non propagation of bosons . in ref .
@xcite it was shown that all spectral dimensions coming from the cutoff based bosonic spectral action do not coincide with the topological @xmath10 , viz .
all of them are zero , implying that some sort of ultraviolet completion , like asymptotic safety @xcite , is necessary .
the @xmath8 spectral action instead exhibits viable spectral dimensions . for higgs scalars and gauge bosons
the spectral dimensions coincide with the topological one and equal four , while in the gravitational sector the spectral dimension equals two , which implies improved ultraviolet behavior of the gravitational propagators .
the @xmath8 spectral action defines naturally , at the classical level , all dimension four operators , therefore the crucial point is the generation of lower dimensional operators , namely the quadratic higgs term , and the einstein - hilbert term if one considers the standard model coupled to gravity .
it turns out , and this is one of the most interesting aspects of the paper , that these terms are generated by the presence , in the dirac operator , of a term corresponding to the right handed neutrino majorana mass .
it is already known @xcite that the presence of a term ( or a field ) of this kind is crucial in order to obtain the experimental higgs mass . for the current study
, this term must contain dimensionful constants , and this leads to the introduction of the needed lower dimensional operators in the spectral action .
compared with the cutoff based approach the zeta spectral action has at the current stage the same predictive power , however the perspective on the hierarchy of scales and the naturalness is conceptually different .
this paper is organized as follows : in sec .
[ sa ] we review the traditional spectral approach and list its successes and open issues . in sec .
[ zsa ] we introduce an altrenative definition based upon the @xmath8 function in an attempt to address and cure some of the drawbacks mentioned in the previous section and discuss in detail the issue of lower dimensional operators and normalization , emphasizing the advantageous aspects of the @xmath8 spectral action .
[ gsd ] is devoted to the spectral dimensions , and sec .
[ last ] contains conclusions and outlook .
in noncommutative geometry and its applications for particle physics the basic object is the spectral triple @xmath11 .
the topology of spaces is described by a ( possibily non commutative ) @xmath12-algebra @xmath13 , represented as operators on a hilbert space @xmath14 of spinors , considered as the matter fields of the theory .
the geometry is encoded in the ( generalized ) dirac operator @xmath15 which contains indiffation on the differential structure of spacetime , and its metric @xcite .
the standard model ( sm ) of strong and electroweak interactions can be explained from a purely geometric approach , considering an `` almost commutative geometry '' , namely the product of an ordinary manifold @xmath16 ( representing spacetime ) times an internal space described by a matrix algebra , with a particular choice of the algebra @xmath13 . to obtain the sm
the minimal choice of @xmath13 is the algebra product of smooth functions on @xmath16 times the finite algebra of complex numbers plus quaternions plus three by three matrices @xcite : a= c^(m)(chm_3(c ) ) .
the dirac operator comprises a continuous part acting on functions of spacetime , times a finite dimensional part which contains the information of the masses and mixings of the physical fermions : d_0=(+ ) i + _ 5d_f , where @xmath17 is the spin connection , @xmath18 is the usual product of the four @xmath19 matrices and @xmath20 is a matrix containing the masses ( or rather yukawa coulings ) of the fermions . the covariant version of the dirac operator is built with the addition of a generic connection , but the real structure @xmath21 , which is otherwise very important for physics , plays no role in this study ,
hence we ignore it to simplify notation . ]
d = d_0+a , where @xmath22 $ ] , with @xmath23 elements of the algebra , is the algebraic representation of the potential connection one - form . in the cutoff approach
the bosonic spectral action reads to the spectral action is needed to differentiate it from the @xmath8 spectral action which we introduce below , and indicate as @xmath24 .
] s_= ( ) , [ bsadef ] with @xmath25 a cutoff function , and the scale @xmath26 is some constant .
this cutoff function is one at the zero value of its argument , and vanishes at @xmath27 .
typical examples considered in the literature are @xmath28 if @xmath29 and zero otherwise , or a smoothened version of it , alternatively one can consider an exponentially decreasing cutoff @xmath30 .
both @xmath25 and @xmath26 are needed to define a finite trace using eigenvalues of the dirac operator , but they have to be considered as inputs . in particular @xmath25 enters in the heat kernel expanded action via its momenta , which are undefined in the theory .
our work starts from the observation that the introduction of a cutoff function and a scale is not the only way in order to construct a finite trace using the dirac operator .
this turns out to be a crucial starting point for the different version of the spectral action , the @xmath8 action we will present in the next section . the spectral action is a classical quantity and can be calculated using the heat kernel expansion .
the result is a polynomial in decreasing powers of the cutoff scale @xmath26 starting from @xmath31 .
the resulting terms depend on the momenta of the cutoff function @xmath25 and the fields which appear in the corresponding powers of @xmath32 , and their derivatives .
the obtained result can schematically be written as @xcite s _ & = & d^4x ( a_1 ^4 .
+ & & + a_2 ^2(54 r - 2 y_t^2h^2 - m^2 ) + & & + a_3 ( g_2 ^ 2w^_w^ + g_3 ^ 2 g^a_g^a + g_1 ^ 2 b_b^ ) + & & + + & & .
+ o(^-2 ) ) , [ sa ] where @xmath33 is the yukawa coupling for the top quark ; @xmath34 is the curvature scalar ; and @xmath35 are the curvature tensors for the three interactions and @xmath16 is ( up to a numerical factor ) a heavy majorana right handed neutrino mass ( which would appear also in the @xmath36 term variously coupled ) .
the @xmath37 with @xmath38 are constants which depend on the details of the function @xmath25 and for typical choices of that function the three constants are not too different from unity . in the case of the cutoff being the characteristic function of the unit interval the @xmath39 are not present in the asymptotic expansion , however careful analysis of the situation shows @xcite that the anzatz coincides with the left hand side only for momenta smaller than the cutoff @xmath26 . in eq .
the @xmath40 , @xmath41 , @xmath42 are the corresponding gauge couplings of the three interactions .
the action we have written is classical , but the relation among the @xmath43 s indicates that the action @xmath44 is written at the scale at which all three gauge constants are equal ( up to the usual factor of @xmath45 normalizing the abelian interaction ) : 53 g_1 ^ 2 ( ) = g_2 ^ 2 ( ) = g_3 ^ 2 ( ) .
[ unif ] present data indicate that this is not true , at least in the absence of new physics , but experimentally it results that the three couplings are very close to each other in the energy range of @xmath46 gev .
modification of the action , as for instance within a supersymmetric theory @xcite , or by considering the six dimensional terms of order @xmath47 @xcite , may lead to a more concrete unification .
the presence of a unification point does not necessarily imply a larger , grand unified group .
in fact the most common scenarios , based on su(5 ) or so(10 ) do not fit in the noncommutative geometry frame , although there is strong indication that some sort of pati - salam symmetry may be present @xcite .
nevertheless , independently of the choice of gauge group among _ allowed _ , the cutoff spectral action requires the presence of an additional scale @xmath48 gev .
the presence of the unification point of the three interactions is fundamental to the theory , and does not depend on the fact that a regularization is needed , in other words it does not depend on the form of @xmath25 or the value of @xmath26 .
its specific value is related to the spectral data contained in @xmath32 , and one might say that its presence is more of kinematical nature .
this observation is important for the alternative definition that we advocate , which exploits the spectral triple data only .
we will first however discuss the successes and open issues of the cutoff bosonic spectral action . from the symmetry point of view
the main point is that very few gauge groups fit into the model .
fortunately the standard model one does , but very few others do .
for example the pati - salam group su(2)@xmath49 su(2)@xmath50 su(4 ) does , but so(10 ) , of which the former group is an intermediate breaking stage , does not
. moreover the formalism allows fermions in the fundamental or trivial representation of the gauge group , a feature of the standard model .
the absence of larger groups with novel representation prevents proton decay .
the main success of the spectral action is however the possibility to infer quantities related to the boson ( and the higgs in particular ) based on the input of only fermionic parameters ( yukawa couplings and mixing ) present in the ( generalized ) dirac operator .
the dirac operator defines also the fermionic part of the action as a structure of bilinear form acting on fermions : s_f = j |d | .
as we said the model has predictive power , and the initial prediction for the mass of the higgs was 170 gev @xcite ; it is now known this number is not correct .
it is nevertheless remarkable that a theory based on mathematical first principles quantitatively predicts a number which is not too far from the experimental one .
taking the input from experiment the models can be improved with the introduction of a scalar field which alters the running of the quartic coupling and makes it compatible with the actual mass of 126 gev @xcite .
this scalar , usually called @xmath51 in this context , had appeared before within noncommutative geometry @xcite as well as in general , see e.g. ref .
although in ref .
@xcite it was inserted ad - hoc in the spectral action principle , there are approaches @xcite , where it comes naturally via the formalism based on introduction of the so called grand symmetry .
alternatively , it is possible a violation of some of the non commutativity conditions , as done in refs .
@xcite . the bosonic spectral action defined by eq . still leaves open some important issues , which we now discuss .
the locality of the theory is far from settled .
although in the low momentum regime the expansion ( [ sa ] ) recovers the standard model action , the high momentum regime does not contain positive powers of the field derivatives @xcite , exhibiting the structure s_~d^4 x ( _ 1 ^2 h _ h^ + _ 2 + _
3a_a^ ) , where @xmath52 and @xmath53 are bosons of spin @xmath54 and @xmath55 respectively ; @xmath56 are constants depending on the particular realization of the model .
the traverse and traceless fluctuations @xmath57 of the metric tensor @xmath58 are defined as follows g _
= _ + , [ gfluct ] where @xmath59 is the planck mass , i.e. they have canonical dimension of energy .
this opens the question of the meaning of the scale @xmath26 , and what happens beyond it . whether the theory is unitary or causal , for example
the theory is certainly not renormalizable , as it stands , since at high momenta the bosonic propagators do not decrease .
for instance , in contrast to conventional qed , the diagram presented in fig .
[ bosonfig ] is divergent , therefore one has to add four fermonic interaction @xmath60 in order to subtract the infinity .
theories with four fermonic interactions are well known to be nonrenormalizable . the spectral action is calculated via the asymptotic heat kernel expansion .
this can be divergent and generally speaking does not coincide with the spectral action @xcite .
let us note that doing a rigorous analysis of the heat kernel expansion one does not actually obtain the higgs potential from the heat expansion , i.e. one does not find a minimum for all natural that equals to 1 at @xmath61 and then rapidly decreases at @xmath62 .
] choices of the cutoff function . indeed ,
although the finite ansatz based on the first three nontrivial heat kernel coefficients reproduces familiar double well potential , the total sum does not .
already for the exponential cutoff the higgs potential defined as follows v(h ) = _ vol , h = const .
= _ , b _ = g
_ = w _ = 0 , can be computed exactly using results of @xcite , and the answer reads v(h ) = _
j=1^n_f b_j , [ vexact ] where @xmath63 and @xmath64 are nonnegative constants and @xmath65 is number of fermmions .
for a non exponential cutoff , we find that performing a laplace transform of the cutoff function , the result remains qualitatively valid for all natural cutoff functions .
the result is valid for all @xmath66 , in particular in the low energy " region @xmath67 where the cutoff spectral action is supposed to work .
therefore , in order to make this model viable it is necessary to add , by hand , to the @xmath68 term , a quadratic term with a large coefficient .
such a term will provide the minimum of the overall potential , while it will lead to the correct higgs vacuum expectation value , which is many order of magnitude smaller than the cutoff scale @xmath26 .
regarding the cosmological constant , let us note that the natural value obtained through the spectral action is @xmath69 , hence much bigger than the observationally known .
hence , one should add by hand the appropriate term to render it compatible with its observational value .
finally it is known , that the coefficient in front of the scalar curvature @xmath34 , that has to be identified with inverse gravitational constant , is ( depending on @xmath26 ) at least one order of magnitude smaller than the value , obtained from experiment , therefore one should also add a corresponding term to the action in order to get the correct value of this coefficient .
thus , we conclude that the magnitude of the _ dimensionful _ parameters appearing in the model , the cosmological constant , the higgs vacuum expectation value and the gravitational coupling have to be put in by hand with unnatural numerical values which are independent of the cutoff scale .
we emphasize that independently of the choice of the almost - commutative manifold , the physical values of these quantities necessitate an experimental input which goes beyond the data encoded by the spectral triple .
all these quantities have to be substituted by a subtraction point which fixes their value by hand to fit the experimental data .
this drawback is closely related to the naturalness problem .
there are other issues , like the signature , or the compactness of spacetime , which are beyond the scope of the present study .
going back to the origins of the bosonic spectral action , one notes that this is a regularized version of the number of eigenvalues of the square of the dirac operator .
the number of eigenvalues of an unbounded operator is of course infinite and one has to ( classically ) regularize this sum , which would be otherwise @xmath70 .
the spectral action does it with the introduction of the cutoff scale @xmath26 .
since the rationale for introducing the spectral action was a regularization , we may try another regularization of the number of the eigenvalues of an operator . to cure some of the drawbacks of the conventional bosonic spectral action outlined above , we propose a definition of the bosonic spectral action based upon the @xmath8 function .
we hence define the bosonic spectral action as s__s0 d^-2s(0,d^2 ) .
[ szetadef ] the @xmath8 function is well defined @xcite and given by the @xmath71 heat kernel coefficient associated with the laplace type operator @xmath72 , namely s_&= & a_4=
d^4 x l , + l(x ) & = & a_4(d^2,x ) .
[ bsav20 ] we refer the reader to e.g. , ref .
@xcite for details . strictly speaking the trace of @xmath73
is convergent if the real part of @xmath74 is greater than the dimension
. then one can prove that it has a unique meromorphic extension , denoted @xmath75 , which has a set of poles on the complex plane of the variable @xmath76 .
for a laplace type operator @xmath72 , the point @xmath77 is not a pole , therefore @xmath78 is well - defined .
the zeta function and its regularity at zero were discussed in @xcite in the context of almost commutative manifolds , and in @xcite in the more general noncommutative setup .
here we use the zeta function to regularize the sum in eq . defining the classical action , while in a slightly different context the zeta function regularization is also commonly used to regularize functional determinants appearing upon quantization @xcite .
the spectral action is nothing but the conformal anomaly in a theory of quantized fermions @xcite where the bosonic fields are a classical background , the relation between the cutoff spectral action and the anomaly can be found in refs .
the lagrangian density obtained from the @xmath8 spectral action has the form : l(x ) & = & _ 1m^4 + _ 2 m^2 r + _ 3m^2h^2 + & & + _ 4 b_b^ + _ 5 w_^w^ + _ 6 g_^ag^a + & & + _ 7h(-^2-)h +
_ 8 h^4 + _ 9 c_c^ + _ 10r^*r^ * , [ l ] where @xmath79 , @xmath80 and @xmath81 are the field strength tensors of the corresponding u(1 ) , su(2 ) and su(3 ) gauge fields ; @xmath82 are dimensionless constants determined by the dirac operator ( whose particular form is not relevant here ) ; @xmath83 is the gauss - bonnet density and @xmath84 is the weyl tensor .
the bosonic spectral action @xmath24 contains only terms needed for the standard model and einstein gravity and _ nothing else _ ( e.g. higher dimensional operators )
therefore it is _ local _ , _ renormalizable _ and _ unitary_. this means that one can use renormalization and safely compute an arbitrary loop order corrections . in this analysis , as in general for the spectral action , gravity is a background field which is _ not quantized _ , therefore there are no issues of renormazability or unitarity as far this sector is concerned .
another strong advantage of the definition is the fact that the lagrangian is an exact result , therefore there is no need to consider asymptotic expansions and their convergence .
substituting the weyl square and gauss - bonnet density expressions via @xmath85 , @xmath86 and @xmath87 we can rewrite our lagrangian as a linear combination & & l(x ) = _
j=1 ^ 12 _ jo_j , where & & o_1 = 1 , o_2 = r , o_3 = h^2 , o_4 = b_b^,o_5 = w_^w^ , + & & o_6 = g_^a g^ a , o_7 = h^2h , o_8 = h^2 r , o_9 = h^4,o_10 = r_r^ , + & & o_11 = r_r^,o_12 = r^2 .
[ genth ] the lagrangian given by eq . is the most general _ renormalizible _ lagrangian for qft in curved spacetime , and correspondingly the complete spectral action s = ||d|+ s _ , is a renormalizable theory describing the standard model .
upon quantization all twelve composite operators @xmath88 in eq . must be renormalized and after proper introduction of the renormalization matrix and counter terms the coefficients @xmath89 by the end of the day must be replaced by renormalized physical parameters @xmath90 .
quantum field theory never predicts the physical values of the coefficients @xmath91 and they must be fixed at some energy scale by normalization conditions . usually such normalization is done using the values obtained from experiment at low energy . for the spectral action
however it is natural to fix the scale at the unification point , and this fixes the relations with all other coefficients , which likewise are normalized at the unification point , _ with their value given by the spectral action_. we emphasize that this normalization procedure is not a consequence of the spectral geometry framework , but is a natural prescription .
this prescription gives predictive power , and we will use it considering the scale at which the @xmath8 spectral action is written to be @xmath92 gev . in analogy with the conventional bosonic spectral action discussed in sec .
[ sa ] , we will still call this scale @xmath26 . in conclusion ,
the bosonic spectral action is written as an action valid at a particular scale , whilst the action is itself independent of this scale .
the spectral approach is very successful giving restrictions on _ dimensionless _ parameters like higgs quartic coupling , gauge couplings and yukawa couplings , etc . for the @xmath9 spectral action in its present formulation
the issue is the value of the dimensionful constants in the lower dimensional terms in the action .
we notice from that the presence of the majorana mass term in the dirac operator introduces the correct lower dimensional operators , however the corresponding coefficients are physically inappropriate . therefore these three numbers can not be taken from the spectral action , and one has to normalize the lower dimensional operators by hand , thereby leaving the naturalness problem unsolved .
let us comment here on the fundamental role played by the dimensionful constant @xmath16 appearing in the position corresponding to the majorana mass in the dirac operator . at this stage
this is a constant quantity , at the end of the paper we comment on the interplay between this term and the @xmath51 field .
as we said , bare values of the dimensionful parameters must be renormalized by hand .
however , the corresponding terms in carry information : they define the structure of the counter terms needed to eliminate divergences upon quantization when one uses dimensional regularization . indeed ,
if one has @xmath93 , since there are no dimensionful constants in the bare lagrangian anymore , divergences proportional to @xmath55 , @xmath34 and @xmath68 would not appear , and there would be no necessity to introduce the corresponding counter terms . correspondingly , the cosmological constant , higgs mass parameter and the gravitational constant would never come out from renormalization . in the context of the spectral action the majorana mass term already plays a fundamental role for the phenomenological viability of the model
; in the present context its role is even enhanced . * remark * _ it is useful to compare our approach with the one of ref .
@xcite where the spectral action was defined by the ansatz of the ( generally divergent ) asymptotic expansion _
n=0^n f_2n ^4 - 2na_2n[d ] , [ aesa ] where @xmath94 are arbitrary and @xmath95 .
this makes the theory local and super - renormalizable , with @xmath26 a cutoff , not a physical scale .
_ the higher terms are a particular kind of higher derivative regularization @xcite , in particular when @xmath96 we have the following action for the gauge field f_4f_^a f^ a + f_^a ( -)^2 f^ a , which improves the ultraviolet behaviour of the propagator . at finite values of @xmath26 such theories
are known to be super renormalizable ( but with ghosts ) and in the limit @xmath97 one recovers the original renormalizable ( without ghosts ) theory .
since there are still divergent one loop fermionic diagrams one would then have to regularize the theory with dimensional regularization , thereby creating an artificial hybrid of higher - derivative and dimensional regularizations @xcite .
for @xmath98 in flat spacetime the action is renormalizable and unitary .
however , the coefficients @xmath99 and @xmath100 that are supposed to introduce the cosmological constant , higgs vacuum expectation value and einstein - hilbert action term do not have by themselves predictive power , since all these parameters have to be normalized using experimental values . if , keeping @xmath101 , one removes them by hand , the definition will lead to our definition .@xmath102
unlike the cutoff formulation our new definition leads to nontrivial spectral dimensions , which we calculate in this section .
the spectral dimension is conventionally the effective dimension of the manifold probed by the particles living " on it ( see for instance ref .
@xcite and references therein for details ) . in particular , for the standard action of a particle with mass @xmath103
, one has to replace @xmath104 with @xmath105 in , leading to a spectral dimension which will coincide with the topological one , namely @xmath106 . for a more complicated momenta dependence
, @xmath107 can be different from the topological dimension , being dependent on the particles one chooses to probe , since their propagators can have different dependence on the momenta .
since the actions for higgs scalar and for gauge fields have the same behavior in the ultraviolet , like in the infrared , their corresponding spectral dimensions coincide with the topological dimension of the manifold and are equal to four .
the gravitational spectral dimension can be also defined in a viable way , however such a definition requires some analytical continuation , therefore we elaborate carefully this point .
the gravitational part of our theory consists of the weyl square contribution coming from eq . and
the ricci scalar @xmath34 appearing after the renormalization discussed in the previous section : s_gr = d^4 x ( r - c_c^ ) , [ sg ] where @xmath108 stands for the number of fermions . to compute the spectral dimension one has to extract the quadratic part of @xmath109 for transverse and traceless fluctuations @xmath57 of the metric tensor @xmath58 defined by .
we obtain s_gr = d^4x h_h^ + o(h^4 ) [ sg2 ] , where a .
to define the spectral dimension one needs the heat kernel @xmath110 corresponding to eq . , or more precisely its value @xmath111 at @xmath112 .
one can see from eq . that such a heat kernel is given by p(t , x , x ) = e^ip(x - x ) e^-(p^2 - a p^4)t .[hk ] note that setting @xmath113 , the integral @xmath111 is divergent , because @xmath114 is positive whilst is well defined for negative @xmath114 . in spherical coordinates the relevant integral is proportional to _ 0^ dp p^3 e^-(p^2 - a p^4)t = , where the right hand side is an analytic function on the complex plane without a ray , that we can choose as a lower half of imaginary axis @xmath115 .
it means that there exists an _
analytic continuation _ in a region of positive @xmath114 ; we _ define _ our integral for positive @xmath114 as such an analytic continuation .
we are now ready to compute the spectral dimension d_s _ t0 .
[ spdim ] following ref .
@xcite , apart from the spectral dimension we also introduce the running " spectral dimension @xmath116 ( t ) -2 , [ runsd ] which in our case is imaginary , however its limit ( i.e. the _ conventional _ spectral dimension ) we will show that is real .
we write ( t)= , and plot @xmath116 in fig . 1 .
although we are interested in the limit @xmath117 , it is worth to note that in the limit @xmath118 the running " spectral dimension is real , while it would be interesting to understand the meaning of the running " imaginary spectral dimension .
returning to the _ conventional _ spectral dimension , we see , that for all _ nonzero _ real @xmath114 we get d_s _ t0 ( t ) = 2 .
[ res ] finally , although in the intermediate range of the parameter @xmath119 the spectral running " dimension is imaginary , there exists a sensible low energy " limit of @xmath120 , valid again for all real @xmath114 , with d_s^low _ t ( t ) = 4 .
[ lowres ] our result in eq .
is quite natural and the fact that the actual spectral dimension is 2 , implies that it is in agreement with the fact , that the gravitational propagators in our theory decrease faster at infinity due to the presence of the fourth derivative , thereby improving the ultraviolet convergence of the feynman loop diagrams . from another point of view our `` low energy ''
result is in agreement with the fact that at very low energies the dynamics does not feel the weyl square terms .
* remark * _ in principle relaxing the normalization condition discussed in the previous section , one can also renormalize the coefficient in front of the weyl square action to a positive constant , that would correspond to negative @xmath114 in eq . .
such a renormalization would decrease a little bit the predictive power for curved spacetime , however all particle phenomenology related with the flat spacetime would remain unchanged
. this normalization may be favorable in order to have positively defined expression ( at large momenta ) for the euclidean path integral , therefore for completeness we present also results for running dimension in this case . _ in this situation , the running spectral dimension @xmath116 is real for all @xmath119 , but not just at @xmath121 and @xmath122 , and the corresponding plot is presented in fig [ figure3 ] .
in this paper we propose a new definition of the bosonic spectral action using the zeta function regularization .
the corresponding theory is local , unitary and renormalizable .
we recall that the mentioned statement concerns fermions and bosons , while quantization of gravity goes beyond the scope of our present study .
such a separation is indeed reasonable ; quantum gravitational effects manifest themselves at energies @xmath123 gev , but nonlocality and nonrenormalizability manifest theirselves at the cutoff scale @xmath124 gev , i.e. at energies _ at least _
100 times less .
the spectral dimensions for fields of various spin are non trivial . in order to obtain ( in the ultraviolet renormalization ) the higgs quadratic term ,
a term in the dirac operator corresponding to a neutrino majorana mass is fundamental .
a nonzero element in that position in the dirac operator is also necessary to obtain the correct mass of the higgs @xcite . in this case
the entry is a field a_i^c ( x ) , i=1,2,3 , [ mast ] ( where @xmath125 is a generation index ) and one can consider more general terms ^c ( a_i(x ) + m_i ) , where @xmath126 and @xmath127 are different constants for right - handed neutrinos in different generations , with the condition @xmath128 , for @xmath129 , otherwise a field redefinition could eliminate them .
these terms are allowed by symmetries .
indeed , the only reason of the usage of constant mass terms was the introduction of @xmath130 , @xmath131 and @xmath132 terms in the action and therefore the corresponding counter terms upon ultraviolet renormalization .
the numerical values of the constants @xmath127 is not relevant , they can be arbitrarily small ( and therefore without phenomenological consequences ) , but they must be nonzero in order to introduce the counterterms needed to renormalize the cosmological constant , the quadratic higgs and the einstein - hilbert terms .
we emphasize , that at the present stage all phenomenological predictions of our approach ( like the prediction of the higgs mass ) are the same as in the ones derived from the cutoff spectral action .
working in the formalism exhibiting both @xmath51 field and the first order condition , e.g. the grand symmetry framework @xcite , then the dirac operator only has neutrino majorana mass terms of the kind of and correspondingly no dimension zero and two operators appear in the classical action . in this case
the @xmath8 spectral action reads : & & s_= dx ( _ 1 b_b^ + _ 2 w_^w^ + _ 3 g_^ag^a + _ 4h(-^2-)h .
+ & & .+ _ 5 h^4+_6(-^2-)+ _ 7 ^4 + _ 8 h^2 ^ 2 + _ 9 c_c^ + _ 10r^*r^ * ) , [ sc ] both fermionic and bosonic parts of the spectral action are invariant under local conformal transformations ( x ) e^-(x ) ( x),h(x ) e^-(x ) h(x),(x ) e^-(x ) ( x ) , e^a_e^+(x)e^a _ , and this classical theory does not contain any dimensionful parameters .
since the theory is renormalizable these parameters will not appear in the renormalization process .
a natural development of the theory described so far is the possibility to generate dynamically the three scales discussed above , thereby predicting them based on the spectral data and the unification point .
dynamical generation of scales upon quantization has a long history dating back from sakharov @xcite for the gravitational sector , and coleman and weinberg @xcite for the higgs sector .
the zeta spectral action is a particular scale invariant extension of the scalar model .
such extensions are promising for the solution of the naturalness ( hierarchy ) problem @xcite . in this approach spontaneous symmetry breaking value is `` triggered '' by quantum correction and the higgs vacuum expectation value can be computed @xcite , thus increasing predictive power .
as far the gravitational sector is concerned , there are several examples where the gravitational constant can be induced , see for example the review @xcite
. it would be fascinating to put a full fledged quantization of gravity in this scheme , nd even if this is still far in the future , we note that possible stating points could be conformal gravity , especially in the bender - mannhiem formalism @xcite , which leads to unitary and renormalizable theory of gravity @xcite and has interesting astrophysical consequences @xcite .
interesting connections between conformal and einstein gravities are discussed in ref .
@xcite . in conclusion , the zeta spectral action is an interesting alternative to the usual cutoff spectral action .
it shows promises of explaining the phenomenology of the standard model and beyond .
in addition , the way it treats the fundamental scales could also shed some light on the explanation of some fundamental questions .
* acknowledgements * we thank o.o .
novikov and p. mannheim for valuable discussions and correspondence , and also f. dandrea and w. van suijlekom for useful comments .
is partially supported by cur generalitat de catalunya under projects fpa2013 - 46570 and 2014 sgr 104 .
m.a.k . and f.l .
were partially supported by unina and compagnia di san paolo under the grant programma star 2013 .
is partially supported by rfbr grant 13 - 02 - 00127-a .
is partially supported by a royal thai government scholarship .
a. d. sakharov , `` vacuum quantum fluctuations in curved space and the theory of gravitation , '' sov . phys .
* 12 * ( 1968 ) 1040 [ dokl .
nauk ser .
* 177 * ( 1967 ) 70 ] [ sov .
* 34 * ( 1991 ) 394 ] [ gen .
rel . grav .
* 32 * ( 2000 ) 365 ] .
|
in this paper we propose a novel definition of the bosonic spectral action using zeta function regularization , in order to address the issues of renormalizability and spectral dimensions .
we compare the zeta spectral action with the usual ( cutoff based ) spectral action and discuss its origin , predictive power , stressing the importance of the issue of the three dimensionful fundamental constants , namely the cosmological constant , the higgs vacuum expectation value , and the gravitational constant .
we emphasize the fundamental role of the neutrino majorana mass term for the structure of the bosonic action . iccub-14 - 068 + kcl - ph - th/2014 - 51 * spectral action with zeta function regularization * maxim a. kurkov@xmath0 , fedele lizzi@xmath1 , mairi sakellariadou@xmath2 , apimook watcharangkool@xmath2 + @xmath3_dipartimento di matematica e applicazioni renato caccioppoli " , + universit di napoli _ federico ii _
_ + @xmath4_infn , sezione di napoli _ + @xmath5_dipartimento di fisica , universit di napoli _ federico
ii _ _ + _ monte s. angelo , via cintia , 80126 napoli , italy _ + @xmath6 _ departament de estructura i constituents de la matria , + institut de cincies del cosmos , universitat de barcelona , + barcelona , catalonia , spain _
+ @xmath7 _ physics department , king s college london , university of london + strand , london wc2 2rls , u.k .
_ 2 cm
|
the rice compression algorithm ( rice _ et al .
_ 1993 ) is particularly familiar to astronomers in combination with the fits tiled image convention ( white _ et al .
_ 2006 , seaman _
it has been informally paired in the past with non - linear data representations ( nieto - santisteban _
_ 1999 , nicula _
mention of compression is often followed immediately by the word `` scheme '' , as if this branch of computer science were suitable only for black - box heuristics arrived at through a process of trial - and - error .
this paper seeks to begin a process of layering the issue of optimal data encoding onto a more formal foundation . as discussed in pence , seaman & white ( 2009a , b ) , the compression achieved for an astronomical image is typically determined almost entirely by its background noise . any processing that quiets
the background will increase the compression ratio .
this is the foundation for all lossy compression algorithms in effect , lossy compression combines a preprocessing step that tempers an image s noise characteristics with the subsequent application of a lossless algorithm .
janesick ( 2001 ) describes a non - linear hardware or software component called a `` square - rooter '' .
instead of using a linear analog - to - digital conversion when reading out a ccd , the square - root of these values acts to linearize the poisson statistics that characterizes these detectors .
such a dn encoding has been used on several space missions to reduce bandwidth requirements for data transport . while technically lossy
if the square - root transform occurs after the a / d conversion , it is equally reasonable to consider this a type of lossless encoding since the transform maintains oversampling of the noise at both the high and low end of the dynamic range .
square - root encoding itself acts as a form of compression
65,636 linear data levels ( for 16-bit pixels ) will turn into a scant 256 levels after the square - root .
this is not nearly as dramatic when expressed in bits , since mapping 16 bits into 8 bits corresponds to a compression ratio of just @xmath0 .
the more important aspect , rather , is to linearize the noise .
a vector of random variables ( _ ie . _ , an image )
is said to be _
heteroscedastic _ if the variance depends on the signal .
many astronomical detectors are governed by photon shot - noise , described by poisson statistics with the noise varying as the square - root of the signal .
bright pixels are noisy pixels .
many familiar statistical techniques such as least - squares fitting formally require _
homoscedasticity_. the penalty for ignoring this requirement for a particular purpose may range from negligible to significant , but
the broader point is that astronomical noise models are often very non - linear .
data compression techniques as applied to astronomical data must take this into account .
techniques exist to stabilize the variance , that is , to convert a heteroscedastic data set to one that is homoscedastic .
one such technique is the anscombe transform ( anscombe 1948 ) : @xmath1 the anscombe transform will convert a poisson sample to have ( nearly ) gaussian statistics .
the factor of 2 ensures a unit variance . real astronomical data does not follow a pure poisson noise model .
for instance , each pixel on a ccd has additive gaussian read - noise as well as poisson shot - noise .
the generalized anscombe transform ( murtagh _ et al .
_ 1995 ) will stabilize the variance in this case : @xmath2 where @xmath3 .
this can be rearranged into a form that uses familiar ccd terminology ( with care to keep units straight between dn and @xmath4 ) : @xmath5 here the quantities have their usual meaning : * @xmath6 is a pixel value in analog - to - digital data numbers ( dn ) * the ccd @xmath7 is in dn * the ccd @xmath8 is in @xmath9 * the readout noise , @xmath10 , is in electrons , @xmath4
in this form , variance stabilization becomes very reminiscent of the ccd photon transfer technique ( janesick 2007 ) .
( many of this paper s conclusions will be applicable to other types of astronomical detectors , particularly in the optical and infrared .
) gain is the slope of a ccd s mean - variance relation , but photon transfer permits much finer grained analysis than simply assuming a scalar gain .
bright pixels greatly oversample the noise .
all digital devices must contend with quantization noise as a term like @xmath11 dn ( widrow & kollr 2008 ) . to avoid stairstep aliasing effects
, the gain of a ccd should be kept lower than the read - noise , typically a value of a few . on the other hand
, the high end of the dynamic range is governed by shot noise , scaling as the square - root of the signal ( in @xmath4 ) . for a gain fixed at unity , the noise for pixels near the top of the 16-bit range will be 256 dn , fully 2 orders of magnitude above the read - noise .
noise adds in quadrature permitting the gain to be scaled with total noise : @xmath12 this is simply another way to look at variance stabilization ( see , for example , figure 11.19 from janesick 2007 ) .
the goal is to tame all the sources of noise .
astronomical images are not simply random agglomerations of pixels , they result from astronomical detectors with specific characteristics and noise models . for ccds ( quantum yield @xmath13 )
this becomes : @xmath14 this presents a quandry , since the fixed pattern noise ( fpn ) dominates both the gaussian read - noise and poisson shot noise at the bright end ( if not flatfielded ) .
the question is how to put this together into a coherent recipe .
work is ongoing , but broad strokes are clear .
first , the quandry is only apparent , since to first order the signal ( and thus the fpn ) is negligible . recall that the compression ratio , _ r _ , is determined by the noise in the _ background _
( pence _ et al .
_ 2009a , b ) : @xmath15 where ( @xmath16 estimated per stoehr 2007 ) : @xmath17 second , phenomenological schemes such as eq . 7.86 from janesick ( 2001 ) : @xmath18 are simplified versions of the more formal functional behavior . equation ( 8) is the same as the anscombe transform when @xmath19 ( _ eg . _ , @xmath20 for _ bitpix _ = 16 ) and when the 3/8 term is negligible ( @xmath21 ) . whatever functional form , a lookup table is an efficient implementation .
the mapping @xmath22 is surjective ; several inputs map to one output .
the inverse lut is compact as a header or table structure under the fits tiled image convention .
libraries like cfitsio ( http://heasarc.gsfc.nasa.gov/fitsio ) that support tiling will be able to use such luts to transparently recover the linear data with adequate noise sampling . for some purposes linearization is not needed .
square root encoding may be just what is wanted to load an image display , or for homoscedastic statistics ( _ eg .
_ , principle components analysis ) , or for multiresolution techniques using wavelets ( murtagh _ et al .
_ 1995 ) . 1 .
since it is the background noise that matters , the low end mapping is key .
however , in equation ( 3 ) each pixel value is corrected for the bias , potentially driving the radicand negative .
how can this best be resolved ? 2 .
what is the best high end mapping for raw and flat - fielded images ? 3 .
when bringing the ccd noise model into alignment with the generalized anscombe transform , it is easy to get lost between @xmath4 and dns ( also , what most call the _ gain _ , janesick calls _ sensitivity _ , the inverse gain ) .
effects like quantum yield need to be folded into the variance stabilization .
4 . the data compression literature ( _ eg .
_ , salomon 2004 ) is built on a foundation of information theory ( _ eg .
_ , cover & thomas 2006 ) . as such , it may be revealing to recast both the variance stabilization and ccd noise modeling techniques in terms of the shannon entropy ( 1948 ) .
anscombe , f. j. 1948 , biometrika , 15 , 246 cover , t. m. & thomas , j. a. 2006,``elements of information theory '' ( wiley - interscience ) janesick , j. 2001 , `` scientific charge coupled devices '' ( spie press ) janesick , j. 2007 , `` photon transfer , dn @xmath23 '' ( spie press ) murtagh , f. , starck , j .- l .
& bijaoui , a. 1995 , , 112 , 179 nicula , b. , berghmans , d. & hochedez , j - f .
2005 , solar physics , 228 , 253 nieto - santisteban , m. et al . 1999 , , 137 pence , w. , seaman , r. & white , r. 2009a , `` lossless astronomical image compression and the effects of noise '' , , in press pence , w. , seaman , r. & white , r. 2009b , , rice , r. f. , yeh , p .- s . & miller , w. h. 1993 , in proc .
9th aiaa comp . in aerospace conf .
, ( aiaa-93 - 4541-cp ) , amer . inst . of aeronautics & astronautics
salomon , d. 2004 , `` data compression : the complete reference , 3rd ed . '' ( springer ) seaman , r. , et al .
2007 , , 483 shannon , c. e. 1948 , `` a mathematical theory of communication '' , bell system technical journal , 27 , 379 & 623 stoehr , f. june 2007 , st - ecf newsletter white , r. et al . 2006 , `` tiled image convention for storing compressed images in fits binary tables '' , http://fits.gsfc.nasa.gov/registry/tilecompression/tilecompression.pdf widrow , b. & kollr , i. 2008 , `` quantization noise '' ( cambridge university press )
|
image compression has been a frequent topic of presentations at adass .
compression is often viewed as just a technique to fit more data into a smaller space .
rather , the packing of data its `` density '' affects every facet of local data handling , long distance data transport , and the end - to - end throughput of workflows . in short ,
compression is one aspect of proper data structuring .
for example , with fits tile compression the efficient representation of data is combined with an expressive logistical paradigm for its manipulation .
a deeper question remains . not just how best to represent the data , but which data to represent .
ccds are linear devices .
what does this mean ?
one thing it does not mean is that the analog - to - digital conversion of pixels must be stored using linear data numbers ( dn ) .
an alternative strategy of using non - linear representations is presented , with one motivation being to magnify the efficiency of numerical compression algorithms such as rice .
|
cryptography is a field of cryptography inclined towards efficient cryptographic implementations , as a response to the demands when using highly constrained hardware in low - cost devices , such as passive rfid - tags and smart cards .
there are trade - offs to consider , e.g. , security , memory and performance .
different constructions appear in different ends in the trade - offs ; for instance , aes and stream ciphers can be implemented efficiently in hardware but do not offer _
provable security_. quite recently , a new trend arose in this area , building cryptographic primitives from problems in learning theory .
problems based on learning theory provide a complexity theoretical foundation , on which the security of the cryptosystem can be based upon .
they also have the property of being easy and efficiently implemented , thereby making them appealing in light - weight cryptography .
being a central problem in learning theory , the lpn problem ( _ learning parity with noise _ ) has shown to be of significance in the field of cryptography .
it is a supposedly hard problem - hard . ] , and is not known to be susceptible to quantum attacks , unlike some other classically hard problems such as factoring and the discrete log problem .
the problem can briefly be described as follows .
let @xmath1 be a @xmath2-dimensional binary vector .
we receive a number @xmath3 of noisy versions of scalar products of @xmath1 from an oracle @xmath4 , and our task is to recover @xmath5 .
let @xmath6 be a vector of length @xmath3 and let @xmath7 . for known random @xmath2 bit vectors @xmath8 , we can easily reconstruct an unknown @xmath9 from @xmath10 using linear algebra . in the lpn problem , however , we receive instead noisy versions of @xmath11 . writing the noise in position @xmath12 as @xmath13 , and assuming each @xmath14 to be small , we obtain @xmath15 in matrix form ,
the same is written as @xmath16 where @xmath17 , and the matrix @xmath18 is formed as @xmath19 .
[ cols="^,>,>,>,>,>,>,^ , > " , ]
we now present numerical results of the improved partial key recovery attack on the authentication protocol lapin @xcite .
the attack we describe concerns the instance of lapin using the degree @xmath20 polynomial @xmath21 and with the parameter @xmath22 . as claimed in @xcite , this given instance is designed to resist the best known attack on ring - lpn within the complexity @xmath23 .
however , we have shown that it is possible to greatly improve the attack complexity .
the improvements can be seen in table [ table : ring - lpn - parameterandsecurity ] . for the distinguishing attack ,
the complexity is only @xmath24 . for actual recovery of the secret polynomial
, we need all five coordinates in the crt representation , which gives an upper bound on the security that is roughly @xmath25 . however , to solve this search problem , we describe a specified approach presented in the appendix that reduces the complexity to about @xmath26 .
a comparison with previous algorithms is given in table [ table : ring - lpn - comparison ] in the introduction .
we have proposed a new generic algorithm to solve the reducible case of ring - lpn . by exploiting the ring structure further ,
our new algorithm is much more efficient than previous algorithms , enough to break the claimed 80-bit security for one of the two proposed instances of lapin .
we have shown that a linear code arising from the crt transform characterizes the performance of this attack through its minimum distance .
this is combined with some standard techniques of using birthday or possibly generalized birthday arguments and efficient recovery through fast walsh - hadamard transform .
the low - weight property of the polynomials in the lapin case makes the problem considerably easier than otherwise and thus makes lapin susceptible to our attack .
using really low - weight irreducible polynomials such as @xmath27 can give rise to linear relations with weight as low as 10 or even less .
we have not seen that such polynomials have been pointed out as very weak before .
the description of the new algorithm was influenced by the lapin case .
there are more improvements that can be described in the general case .
one such improvement is the use of a generalized birthday technique @xcite .
this will allow us to consider larger dimensions at the cost of increasing the noise level .
we have also noted that the simple bit - oriented samples in this paper can be replaced by more complicated vectorial samples , which will give a stronger bias .
in this section , we describe the lapin two - round authentication protocol .
let @xmath28/(f)$ ] be a ring and @xmath29 the set of units in @xmath30 .
the protocol is defined over the ring @xmath30 .
let @xmath31 be a mapping , chosen such that for all @xmath32 , @xmath33 if and only if @xmath34 .
furthermore , let @xmath35 be a bernoulli distribution parameter and @xmath36 a threshold parameter .
the elements @xmath30 , @xmath37 , @xmath38 and @xmath39 are public parameters .
the ring elements @xmath40 constitute the secret key .
suppose that we have a key - generation oracle ; it will give us the key @xmath41 .
the secret key is shared among the tag and the reader .
protocol [ prot : ring - lpn - lapin ] gives how information is exchanged between the tag and the reader in lapin .
in this section , we describe a better secret recovering approach on the instance proposed in lapin with a reducible polynomial .
this attack exploits the secret polynomial s representations in the three relatively easy - attacked quotient rings @xmath42/(f_i ) , i = 1 , 2 , 3 $ ] , to decrypt those in rings @xmath42/(f_i ) , i = 4 , 5 $ ] , and thus obtains higher efficiency than simply attacking one ring by another .
actually , it reduces the attacking complexity from @xmath25 to @xmath26 .
we first recover the secret polynomials @xmath48 , @xmath49 and @xmath50 in the corresponding smaller rings .
this step costs around @xmath26 bit operations .
then , using these known polynomials , we show that the secrets @xmath51 and @xmath52 can be recovered with negligible costs .
the second step is transforming each ring - lpn sample to its corresponding standard lpn samples .
this step will be tricky as we want to make use of the known information . since @xmath53 are distinct irreducible polynomials over @xmath42 $ ] , there exist polynomials @xmath54 such that @xmath55 and @xmath56 , where @xmath57 .
moreover , these polynomials @xmath58 can be computed efficiently .
then , for each sample ( @xmath59 ) , where @xmath60 , by adding the polynomial @xmath61 to @xmath62 , we create a new polynomial @xmath63 , which will be converted to a vector @xmath64 , where @xmath65 is the matrix whose @xmath12-th column is the coefficient vector of @xmath66 .
the degree of @xmath51(@xmath52 ) is @xmath67(@xmath68 ) , thereby yielding that only the first @xmath67(@xmath68 ) columns of the matrix @xmath69 , denoted @xmath70 , are useful .
we can compute those sub - matrices in @xmath71 bit operations for each ring - lpn sample , thereby economically constructing @xmath20 standard lpn samples @xmath72 for @xmath73 , where @xmath74 is the @xmath12-th row in the matrix @xmath75 .
the last step is to attack a standard lpn problem with length @xmath76 and error probability @xmath77 . by levieil - fouque algorithm
, the cost is small , i.e. , around @xmath78 bit operations .
it requires @xmath79 standard lpn samples , i.e. @xmath80 ring - lpn ones .
the number of samples in the second step is also bounded by that required in the final step ; the overall cost of the last two steps , therefore , is no more than @xmath81 bit operations , far less than that required in the starting step .
thus , it is reasonable to embrace @xmath26 bit operations as the total attacking complexity .
fossorier , m.p.c . ,
mihaljevic , m.j . ,
imai , h. , cui , y. , matsuura , k. : a novel algorithm for solving the lpn problem and its application to security evaluation of the hb protocol for rfid authentication .
cryptology eprint archive , report 2012/197 ( 2012 ) , http://eprint.iacr.org/ gilbert , h. , robshaw , m.j.b .
, seurin , y. : hb@xmath82 : increasing the security and the efficiency of hb@xmath83 . in : smart , n.p .
eurocrypt 2008 .
lncs , vol .
4965 , pp .
springer , heidelberg ( 2008 ) gilbert , h. , robshaw , m.j.b . ,
sibert , h. : an active attack against hb@xmath83a provably secure lightweight authentication protocol .
cryptology eprint archive , report 2005/237 ( 2005 ) , http://eprint.iacr.org/ kiltz , e. , pietrzak , k. , cash , d. , jain , a. , venturi , d. : efficient authentication from hard learning problems . in : paterson ,
( ed . ) eurocrypt 2011 .
lncs , vol . 6632 , pp .
springer , heidelberg ( 2011 ) lyubashevsky , v. : the parity problem in the presence of noise , decoding random linear codes , and the subset sum problem , in : chekuri , c. , jansen , k. , rolim , j.d.p . ,
trevisan , l. ( eds . ) approx - random 2005 , lncs , vol . 3624 , pp .
378389 . springer , heidelberg ( 2005 )
|
the lpn ( learning parity with noise ) problem has recently proved to be of great importance in cryptology .
a special and very useful case is the ring - lpn problem , which typically provides improved efficiency in the constructed cryptographic primitive .
we present a new algorithm for solving the ring - lpn problem in the case when the polynomial used is reducible .
it greatly outperforms previous algorithms for solving this problem . using the algorithm
, we can break the lapin authentication protocol for the proposed instance using a reducible polynomial , in about @xmath0 bit operations .
birthday attacks , fast walsh - hadamard transform , lapin , lpn , ring - lpn .
|
since the theoretical prediction of magneto - phonon resonance in graphene and bilayer graphene @xcite , the resonant interaction between inter - landau level electronic excitations and @xmath3-point optical phonon in graphene , this effect has been widely explored experimentally in neutral graphene - like systems such as multi layer epitaxial graphene on c - face sic @xcite , graphene like inclusions on the surface of bulk graphite @xcite , and in bulk graphite @xcite .
magneto - phonon resonance in these systems are pronounced because ( i ) they are gapless and low energy direct ( @xmath4 ) electronic excitations exist , and ( ii ) because of the kohn anomaly @xcite at the @xmath3 point of the phonon brillouin zone which makes phonon energies _ ultra sensitive _ to modifications of the electronic excitation spectrum .
such modifications can be achieved by tuning the fermi energy or by applying a magnetic field perpendicular to the plane of the quasi-2d crystal . performed in systems with a well - known electronic excitation spectrum ,
magneto - phonon resonance experiments provide precise information concerning the electron - phonon interaction through the amplitude of the observed oscillations of the energy and of the line width of the phonon feature observed in raman scattering experiments , as a function of the applied magnetic field . on the other hand , since electronic excitations which couple to the e@xmath1 optical phonons are well identified by symmetry arguments ( optical - like excitations characterized by a change of angular momentum @xmath5
@xcite , where @xmath6 is the landau level index ) , this effect can be used to determine the electronic excitation spectrum . the evolution of optical - like excitations , calculated in the frame of ref .
@xcite , as a function of the magnetic field is presented in figure [ fig1 ] for a ) graphene , b ) bilayer graphene , c ) tri - layer graphene and d ) for four - layer graphene .
each of these distinct electronic excitation spectra are tuned in resonance with the phonon energy at different characteristic values of the magnetic field . in analogy to magneto - transport experiments which
probe the density of states at the fermi energy , magneto - phonon resonance can be used as a probe of @xmath5 electronic excitation at the e@xmath1 phonon energy , and landau level spectroscopy can be achieved by modelling the observed shifts of the phonon feature .
evolution of inter landau level excitations with @xmath7 calculated in the frame of the model developed in ref .
@xcite , in the case of ( a ) a monolayer graphene , ( b ) a bilayer graphene , ( c ) a trilayer graphene and ( d ) four - layer graphene .
the values of @xmath8 used in these calculations are indicated in the figures .
the phonon energy is indicated by the solid horizontal line at 1585 @xmath9 .
dashed vertical lines indicate resonant magnetic fields corresponding to the different excitation spectra .
colors are only used to differentiate the different excitation spectra.,scaledwidth=100.0% ] in this paper , we report on low - temperature magneto - raman scattering experiments revealing the magneto - phonon resonance of an exfoliated multi - layer graphene specimen deposited on a sio@xmath2/si substrate .
the investigated sample is characterized by a low `` intrinsic '' doping and oscillations of both the phonon energy and line width can be observed for magnetic fields @xmath10 .
these oscillations have a noticeably more complicated pattern than for monolayer graphene and for bulk graphite . from the analysis of this pattern we deduce the number of layers in the sample , which turns out to be equal to four .
observation of a magneto - phonon resonance can hence be used to trace electronic excitations at the phonon energy and determine the band structure of the probed electronic system .
the multilayer graphene specimen is presented in the inset of figure [ fig2]a .
it has been exfoliated from bulk graphite and deposited on a si / sio@xmath2 substrate with @xmath11 nm of sio@xmath2 .
a single metallic electrode was then deposited on the flake ( not used in this experiment ) . to locate the sample , we spatially map the raman scattering signal of the silicon substrate , which shows a strong attenuation when the laser spot is focused on the metallic layer .
a typical room temperature raman scattering spectrum of the multilayer graphene flake measured with labram confocal micro - raman scattering set - up , together with an optical image are presented in figure [ fig2]a .
the 2d band has a multi - components broad feature .
we find the standard analysis of 2d band features in multi - layers specimens , that is , the deconvolution of the complex 2d band line shape using @xmath12 lorentzian functions ( where @xmath13 is the number of layers ) , relevant for mono - layers , for bi - layers and for tri - layers @xcite , not precise enough for @xmath14 . as a consequence
, 4-layer graphene flakes can hardly be distinguished from the 5-layer and additional information is needed to fully characterized multilayer graphene systems .
a ) raman scattering spectrum of the four - layer graphene specimen measured at room temperature with a labram confocal micro - raman scattering set - up , with the @xmath15 nm line of an argon laser .
inset : microscope image of the 4-layer graphene flake .
b ) schematics of the micro - raman scattering set - up operating at liquid helium temperature and in high magnetic fields.,scaledwidth=100.0% ] the low - temperature magneto - raman scattering response of our multilayer graphene has been measured with a home - made micro - raman scattering setup presented in figure [ fig2]b .
laser excitation at @xmath16 produced by an ar ion laser is injected into a @xmath17 core mono - mode optical fiber which is fixed on a miniaturized optical bench and placed at the center of the magnet .
this experimental setup can host various optical filters which are used to filter the excitation laser light at the output of the mono mode fiber ( laser line filter ) , to selectively reflect the excitation laser and transmit the scattered signal ( dichroic mirror ) , to reject the scattered laser before the collection fiber ( notch filter ) , and to impose a circular polarization ( linear polarizers and quarter wave plate ) .
the excitation spot on the surface of the sample has a diameter of @xmath18 and the optical power is of 5 mw .
the sample is mounted on piezzo stages which allow to move the sample under the laser spot with a sub - micrometer resolution .
polarized raman scattering spectra were measured in nearly back - scattering faraday geometry .
the collected light was dispersed with a single grating spectrometer ( spectral resolution @xmath19 ) equipped with nitrogen cooled ccd detector and band pass filters were used to reject the stray light .
in order to obtain further information concerning our multilayer graphene sample , we have performed magneto - raman scattering experiments in the g - band range of energy . the g - band feature of the raman scattering spectrum is observable in the crossed circular polarization configuration @xcite which is used in the present experiment .
even though electronic excitations can not be , in the present case , directly imaged on the raman scattering spectra , their effect when tuned in resonance to the e@xmath1 optical phonon by applying a magnetic field can be clearly observed . false color map of the scattered intensity in the energy range of the @xmath20 phonon as a function of the magnetic field in the @xmath21 ( a ) and @xmath22 ( b ) polarization configurations.,scaledwidth=100.0% ] a ) raman scattering spectra measured in the @xmath21 polarization configuration at different values of the magnetic field .
the dashed vertical line indicates the b=0 energy of the @xmath20 phonon .
b ) intensity of the scattered light at the phonon energy at b=0 as a function 1/b .
c ) amplitude of the fourier transform of b ) showing two characteristic frequencies.,scaledwidth=100.0% ] the magneto - phonon resonance of our multilayer graphene sample is presented in figure [ fig3 ] in the form of false color map of the scattered intensity as a function of the magnetic field , in the two crossed circular polarization configurations @xmath23 .
individual spectra at characteristic values of magnetic field in @xmath24 configuration are presented in figure [ fig4]a . in both configurations ,
we observe , for @xmath10 , oscillations of the energy of the @xmath20 phonon and of its line width .
this low value of the magnetic field for the onset of magneto - oscillations is an indication that the carrier density in this sample is lower than half of the @xmath20 phonon energy .
the resonant values of the magnetic field are different in the two polarization configurations , @xmath25 in the @xmath26 configuration and @xmath27 in the @xmath24 configuration
. moreover , the electron - phonon coupling appears to be more efficient at particular values of the magnetic field ( indicated above by the bold typeface ) where the amplitude of the observed oscillation is enhanced with respect to other resonant magnetic fields .
this nontrivial behavior is different from the one observed for epitaxial graphene , for graphene - like domains on the surface of bulk graphite or for k - point carriers in bulk graphite , where the amplitude of the oscillations was found to increase with the magnetic field as the effective electron - phonon coupling strength depends on the degeneracy of the landau levels which increases linearly with @xmath28 . to determine the number of landau level fan charts contributing to the observed magneto - phonon resonance
, we extract from the spectra presented in figure [ fig3]a and in figure [ fig4]a the evolution of the amplitude of the scattered light at the phonon energy at @xmath29 as a function of the inverse magnetic field .
this quantity shows oscillations which are representative of the phonon oscillations : each time the phonon feature splits or is shifted due to the resonant interaction with an electronic excitation , the amplitude of the scattered light at the phonon energy at @xmath29 decreases .
this evolution is presented in figure [ fig4]b .
it is then possible to numerically calculate the fourier transform of this signal , presented in figure [ fig4]c , which shows two distinct frequencies at about 21 and 49 t , corresponding to two periods @xmath30 of about @xmath31 and @xmath32 .
their appearance can be understood qualitatively using the results of ref .
@xcite for the electronic spectrum of a @xmath33-layer graphene .
namely , the spectrum is similar to that of a bulk graphite , but instead of the continuous quantum number @xmath8 , @xmath34 ( the wave vector , perpendicular to the layers , in the units of the inverse interlayer spacing ) , only @xmath35 ( for @xmath33 even ) or @xmath36 ( for @xmath33 odd ) discrete values , @xmath37 or @xmath38 , respectively , are allowed .
now , instead of a continuum of inter - landau - level excitations for all @xmath8 in graphite , which couple to the phonon @xcite , in the 4-layer graphene one has two series of inter - landau - level excitations , corresponding to the two allowed values of @xmath8 .
thus , the presence of two periods indicated that our sample has either three or four layers . in the two - band approximation for the effective bilayer model of ref .
@xcite , the values of the magnetic field @xmath39 at which the magnetophonon resonance occurs , are determined by @xmath40 where @xmath41 is the fermi velocity in the monolayer , and @xmath42 is the nearest - neighbor inter - layer matrix element .
approximating @xmath43 , we obtain @xmath44 the coefficient in front of @xmath6 thus gives the period @xmath30 for each value of @xmath8 .
for the monolayer - type spectrum at @xmath45 , the same consideration ( again , approximating @xmath43 ) gives @xmath46 for @xmath47 and @xmath48 solving this equation brings an inverse period of 7.1 t , which is too small compared to the smallest observed 21 t. thus , we exclude the possibility that our sample could be a trilayer graphene . for four layers
, we have @xmath49 , which , for @xmath50 , gives two frequencies of 16 and 41 t. this is an indication that the sample has four layers . to give more solid arguments , we perform a quantitative modelling of the spectra .
the raman scattering spectra presented in figure [ fig4 ] can be described by lorentzian functions to extract the energy corresponding to the maximum of scattered light and the full width at half maximum ( fwhm ) of the phonon feature .
these results are presented by the black dots in figure [ fig5 ] ( a - d ) for both polarization configurations .
again , the observed behavior is not a simple series of oscillations ; rather , two series of oscillations can be distinguished .
qualitatively , this is explained by the same arguments as above ( two allowed values of @xmath8 ) .
the theoretical curves are shown in figure .
[ fig5 ] as solid curves on top of the experimental symbols .
evolution of the maximum of scattered light and of the fwhm of the g band as a function of the magnetic field in the @xmath21 ( a and b ) and in the @xmath22 polarization configurations .
red solid lines are the result of the calculation detailed in the text.,scaledwidth=100.0% ]
in contrast to bulk graphite , where due to the translational invariance in the @xmath51 direction , the momentum conservation holds , which implies that only phonons with the wave vector @xmath52 are probed ( assuming the photon momentum to be very small ) , and that the electronic transitions involved are those from states with some @xmath8 into states with @xmath53 , the 4-layer graphene lacks the translational invariance in the @xmath51 direction , which introduces some additional difficulties in the theoretical description of the magneto - phonon resonance , as compared to the bulk graphite .
( i ) there is _ a priori _ no reason to focus on one particular phonon mode . in total , there are eight degenerate optical phonon modes with zero in - plane wave vector ( two for each layer ) .
the circular basis for the in - plane polarizations is still a good one , thus for each circular polarization there are four phonon modes .
we assume that their degeneracy is lifted by the electron - phonon interaction , neglecting the elastic coupling between the layers .
thus , for each of the two circular polarizations , the normal modes should be found as eigenvectors of the electronic polarization operator @xmath54 , which is a @xmath55 matrix .
( ii ) the electronic polarization operator not only has a contribution from the @xmath8-diagonal electronic transitions , but also from transitions between states with different @xmath8 , since no momentum conservation holds .
this gives not just two series of transitions , but four , corresponding to all possible combinations of @xmath56 .
still , it can be naturally expected ( and is confirmed by the actual calculation ) that the overlap between states with different @xmath8 is much smaller than that between states with the same @xmath8 ; this is why two series of oscillations instead of four are seen in figure [ fig4 ] .
( iii ) generally speaking , all four phonon modes should contribute to the raman spectrum .
the intensities of the four components are determined by the projections of the corresponding mode eigenvectors on the raman matrix element ( photon - photon - phonon vertex ) , which is also a four - component vector for each given circular polarization configuration . in bulk graphite , due to the momentum conservation
, the raman vertex would be simply proportional to a @xmath57-function @xmath58 , but in the 4-layer sample it has a non - trivial structure .
the electronic band structure of multilayer graphene is modelled similarly to koshino and ando @xcite .
namely , in the nearest - layer tight - binding approximation , the band structure of any bernal - stacked multi - layer graphene can be deduced from the one of the bulk graphite by selecting specific values of @xmath8 corresponding to the possible standing waves perpendicular to the basal plane of the multilayer graphene specimen .
the monolayer graphene then corresponds to a single value of @xmath45 , the bilayer to @xmath59 , the trilayer corresponds to two values of @xmath60 , etc . at each value of @xmath8
, the hamiltonian has the form of an effective graphene bilayer : @xmath61,\ ] ] where @xmath62 @xmath63 , @xmath64 is the distance between the neighboring carbon atoms in the same layer , @xmath65 are the in - plane momenta counted from the corners of the hexagonal first brillouin zone , and @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 are the parameters of the tight - binding slonczewski - weiss - mcclure ( swm ) model @xcite , whose values we take the same as in ref . @xcite . as in ref .
@xcite , we have neglected the @xmath72 term , responsible for the trigonal warping , in order to simplify the diagonalization of the electronic hamiltonian with an external magnetic field .
the @xmath73 terms in the swm model correspond to the second - nearest - layer coupling , and their non zero values modify the electronic wave functions on the layers with respect to the simple recipe of ref .
@xcite . here
, we consider the wave functions of ref .
@xcite but we still include the @xmath74 and @xmath75 terms in eq .
( [ bilayerham= ] ) , since the experimental transition energies are better reproduced when also considering these terms .
thus , even though eq .
( [ bilayerham= ] ) , strictly speaking , does not correspond to the swm model , we use it as an effective electronic hamiltonian for the 4-layer graphene . given eq .
( [ bilayerham= ] ) , the calculation of the polarization operator is analogous to the one presented in ref .
@xcite with summation over @xmath76 instead of integration over @xmath8 . following points ( i ) and ( ii ) discussed at the beginning of this section
, we also introduce a layer index for the electron - phonon coupling .
the solid curves presented in figure . [ fig5 ] were obtained using the values @xmath77 , @xmath78 for the dimensionless electron - phonon coupling constant and the electronic broadening respectively , but the values @xmath79 , @xmath80 or @xmath81 , @xmath82 differ very little .
( we remind that the main purpose of the present modelling is not to extract precise values of parameters but to demonstrate that the sample is quadri - layer graphene . ) the third difficulty , mentioned at the beginning of this section , can be bypassed since we are interested only in the positions and widths of the raman peaks , determined by the complex eigenvalues of the electronic polarization operator @xmath54 , i. e. , by the complex roots of the equation @xmath83=0,\ ] ] and not in their intensities .
moreover , it turns out that the four eigenvalues behave quite differently upon varying the magnetic field .
only one eigenvalue exhibits pronounced magneto - oscillations , while other three remain close to the bare phonon frequency .
this behavior can be understood by noting that near each resonance @xmath84 , the polarization operator can be approximated as @xmath85 , where @xmath86 and @xmath87 are the frequency and the broadening of the corresponding transition , and @xmath88 is the corresponding matrix element of the electron - phonon coupling ( the indices @xmath89 label the layers where the phonon displacements occur ) .
then , near the resonance , one phonon eigenvector is along @xmath90 , and other three are orthogonal to it , so that they are effectively decoupled from the resonance .
the theoretical curves in figure [ fig5 ] correspond to the eigenvalue exhibiting the non - trivial oscillating behavior .
nevertheless , the contribution of other modes , not subject to oscillations , can also be distinguished in the spectrum of fig .
[ fig3]a ) around the strongest resonance at @xmath91 .
finally , we note that the possibility to avoid the calculation of the raman matrix element in the 4-layer structure is quite fortunate , since its calculation requires a significant effort even in the monolayer graphene @xcite .
we have shown that magneto - phonon resonance in multi - layer graphene specimens can be used to determine the specific band structure of these systems .
we demonstrate this possibility in this work with a 4-layer graphene flake deposited on sio@xmath2 .
magneto - phonon resonance in this sample result in a series of oscillations of the phonon energy which can be directly related to resonances between two distinct landau levels fan charts , characteristic of the 4-layer graphene system .
part of this work has been supported by ec mtkd - ct-2005 - 029671 , euromagnet ii under the eu contract number 228043 and pics-4340 projects .
is financially supported by the eu under fp7 , contract no .
221515 ` mocna ' .
|
we show how the magneto - phonon resonance , particularly pronounced in sp@xmath0 carbon allotropes , can be used as a tool to probe the band structure of multilayer graphene specimens . even when electronic excitations can not be directly observed , their coupling to the e@xmath1 phonon leads to pronounced oscillations of the phonon feature observed through raman scattering experiments with multiple periods and amplitudes detemined by the electronic excitation spectrum .
such experiment and analysis have been performed up to 28 t on an exfoliated 4-layer graphene specimen deposited on sio@xmath2 , and the observed oscillations correspond to the specific ab stacked 4-layer graphene electronic excitation spectrum .
|
when considering how carbon could be produced in stellar nucleosynthesis , hoyle @xcite famously predicted that the @xmath1c nucleus must have an excited @xmath0 state in the vicinity of the 3-@xmath3 breakup threshold .
such a state was duly found @xcite in short order on the basis of hoyle s suggestion .
its existence was essential for @xmath1c to be produced at an adequate rate to open up the pathways to the synthesis of still heavier nuclei @xcite .
not only had this state escaped both experimental detection and theoretical prediction up to that point , but it has continued to pose severe challenges to nuclear structure models ever since .
the shell model struggles to describe any low - lying excited @xmath0 states in p - shell and sd - shell nuclei . to this day
, the most advanced no - core shell model has not succeeded in reproducing the excitation energy of the hoyle state @xcite .
models invoking a 3@xmath3 chain @xcite were able to accommodate the excitation energy , but could not reproduce the decay width of the hoyle state .
better success was enjoyed by orthogonality condition model calculations of the 3@xmath3 system , using a semi - microscopic @xmath8 interaction @xcite , which gave a suitably sized @xmath3-decay width in the @xmath2be@xmath9 channel .
this suggested that the dominant structure of the hoyle state must be @xmath2be@xmath9 in a relative s - state .
such a conclusion was backed up by fully microscopic @xmath10 calculations using the resonating group and generator coordinate methods @xcite .
more recently fully microscopic calculations using antisymmetrized molecular dynamics ( amd ) @xcite and fermionic molecular dynamics ( fmd ) @xcite have been able to give a good account of the low - lying spectrum of @xmath1c without assuming alpha clustering _ a priori_. for the particular case of the very loosely bound hoyle state
a three-@xmath3 condensate wave function @xcite has also been shown to have a large overlap with the fmd wave function .
the current situation is well summarized in a recent review @xcite . in parallel with these theoretical developments ,
new life has recently been breathed into the experimental program .
given that the hoyle state has a large mean square radius , it is natural to suggest calculating higher angular momentum states with this same structure thereby producing a band of excited states with a large moment of inertia .
although friedrich et al .
@xcite claim there are no such rotational states , most @xmath10 models do support them , with the @xmath4 state generally expected to have an excitation energy in the region of 10 mev .
initial searches for this @xmath4 state via beta decay @xcite were discouraging , but in recent times three separate experiments have provided evidence of its existence .
itoh et al .
@xcite see a @xmath4 state at 9.9 mev , with a width of 1.0 mev , via inelastic @xmath3-particle scattering in their @xmath1c(@xmath11 ) measurements .
freer et al .
@xcite see a @xmath4 state at 9.6 mev , with a width of 0.6 mev , in their inelastic proton scattering @xmath1c(p , p@xmath12 ) work .
gai @xcite confirms the existence of a @xmath4 state around 10 mev ( but without being able to measure a width ) in photonuclear disintegration @xmath1c(@xmath13 ) studies . in view of this resurgent interest in an excited hoyle band
we ask in this paper to what extent the known data can be accounted for ( and further excited states predicted ) by a local potential @xmath2be+@xmath3 cluster model .
this model provides a physically transparent and calculationally straightforward description of the energies , widths and charge radii of the known states in the proposed hoyle band .
it also throws some light on the question of how high in angular momentum such a band might continue . in this way it can be a useful guide to experimental groups searching for as yet unidentified higher @xmath14 states .
it is also illuminating to see how far one can get with a simple but physically motivated model .
although ideally the full rigours of a more microscopic and computationally intensive approach might seem preferable , a rather simplified effective nucleon - nucleon potential is needed to carry this to fruition .
it might be that a phenomenological approach can produce a band of states with properties closer to the experimental values . in any event
it is interesting to compare the results from different theoretical treatments with each other as well as with experiment . in the next section
we describe our local potential @xmath2be+@xmath3 cluster model .
after that , we compare the model s results with the available data and discuss the possible existence of additional states .
finally , we summarize our conclusions .
the cluster model employed here was first proposed to study the excited 4p-4h band of @xmath15o ( bandhead at 6.05 mev ) and the ground state band of @xmath16ne @xcite and has subsequently been applied to a wide range of nuclei across the periodic table from @xmath17li to @xmath18 cm .
however , it has not previously been used for @xmath1c , because the ground state band of that nucleus is known to be oblate ( from the sign of its quadrupole moment @xcite ) and a two - body cluster - core system inevitably produces a prolate deformation .
this is because , for spinless constituents like @xmath2be and @xmath3 , the quadrupole operator for the system reduces to @xmath19 the expectation value of this operator always yields a positive quadrupole moment , indicating a prolate spheroidal shape . to obtain negative quadrupole moments , appropriate to oblate shapes
, it is necessary to consider three - body systems .
this outcome provides a large part of the motivation behind attempts to model the ground state band of @xmath1c ( and indeed the excited @xmath20 state at 9.64 mev ) as an equilateral triangular arrangement of three @xmath3 particles @xcite .
however , the indications that the hoyle state , and any associated rotations of it , have a large @xmath2be@xmath9 component , mark out the hoyle band as ideal territory for our local potential two - body cluster model .
unfortunately , this does mean that we are unable to address any states of the @xmath1c system that do not have this particular structure .
in particular , we can not model the ground state band and the excited @xmath20 state mentioned above .
in general , we model a nucleus as two even - even sub - nuclei of mass @xmath21 and @xmath22 separated by a distance @xmath23 , interacting through a deep , local nuclear potential @xmath24 and a coulomb potential @xmath25 appropriate to a point cluster and a uniformly charged spherical core .
the nuclear part has previously been parametrised in the form @xcite : @xmath26 } \nonumber \\\ & + & { 1-x \over [ 1+\exp{((r - r_0)/3a ) } ] ^3 } \biggr \}\end{aligned}\ ] ] with parameter values given by @xmath27 and @xmath28 related to @xmath29 by @xmath30 } + { 1-x \over [ 1+\exp{(-r_0/3a ) } ] ^3 } } } .\ ] ] the radius parameter @xmath31 of the potential is determined by fitting to the experimental energy of the spectrum s band head , and is linked to the choice of @xmath5 ( see below ) .
this potential is now incorporated into the cluster - core relative motion hamiltonian @xmath32 , and the resulting schrdinger equation @xmath33 is solved numerically .
we label the wave functions and energies with the global quantum number @xmath5 , where @xmath34 is the number of internal nodes in the radial wave function and @xmath14 the orbital angular momentum .
we must choose the value of @xmath35 large enough to guarantee that the pauli exclusion principle is satisfied by excluding the cluster constituents from states occupied by the core nucleons .
for the @xmath2be + @xmath3 system this requires @xmath36 .
the appropriate choice of @xmath35 is not clear cut when we are not using an oscillator potential , although we can certainly use oscillator considerations as a guide .
the lowest allowed value , @xmath37 , would correspond to packing the @xmath3 nucleons into the p - shell , which seems an unlikely description of an excited state in @xmath1c .
nevertheless , for completeness , we present calculations employing @xmath38 and 8 , with a view to deciding _ a posteriori _ on the basis of a comparison with experimental data which is the best choice .
indeed , only _ a posteriori _ can we say that our model , with any eventually preferred value of @xmath35 , is appropriate for describing the hoyle state and its associated band by comparing its predictions with the measured properties of those states .
although numerical solution of the schrdinger equation without any restrictions on @xmath35 certainly does produce lower lying states , they are pauli forbidden and do not correspond to physical states of @xmath1c .
thus , we have no @xmath1c bound states in our model and can not use it to describe the observed ground @xmath0 or excited @xmath4 ( 4.44 mev ) states of the nucleus .
equally , we can not describe the @xmath20 ( 9.64 mev ) state of @xmath1c ( as explained earlier ) .
it is possible to produce negative parity states in our model by solving the schrdinger equation with an odd value of @xmath35 , but the resulting energies are 1020 mev above the hoyle state , and not of immediate interest .
similarly , we could obtain negative parity states of @xmath1c by using even @xmath35 values in conjunction with an excited negative parity state of the @xmath2be core .
again , the resulting excitation energies in @xmath1c are too high to be of interest in the current study . as a further model extension
we could include excitations of the @xmath2be core into its @xmath4 and @xmath39 states .
we do not do this here because it would involve the introduction of more adjustable parameters to describe the non - central interactions that accompany such excitations , and we are trying to keep the number of fitted quantities to a minimum .
our previous experience of including core excitations in treatments of @xmath15o @xcite and @xmath40 mg @xcite leads us to expect that they would not have a major effect on our conclusions in the the @xmath2be@xmath41 case .
solving the schrdinger equation produces excitation energies and their associated widths directly .
the resulting wave functions can also be used to calculate mean square charge radii of the states from @xmath42 where @xmath43 and @xmath44 for the @xmath2be + @xmath3 system . to apply this formula for charge radii we need to supplement the wave functions generated above with a description of the ground state of @xmath2be ( or at least , of its mean square charge radius ) .
an excellent description of @xmath8 scattering phase shifts and @xmath2be has already been given within the local potential cluster model @xcite using a gaussian nuclear potential @xmath45 with @xmath46 mev and @xmath47 @xmath48 and a coulomb potential @xmath49 with @xmath50 and where @xmath51 for the @xmath8 system .
we adopt this description of @xmath2be wholesale because , within our model , it can hardly be improved upon . as a consistency check on the widths of the states
we can make use of our earlier work ( see for example @xcite ) on charged particle decay widths . within the two - body local potential model
, a semiclassical approximation leads to an @xmath3-partial width of @xmath52 where @xmath53 and @xmath54 are the two outermost classical turning points and @xmath55 is the reduced mass of the system .
the normalization factor @xmath56 is given to good accuracy by @xmath57 with @xmath58 the innermost turning point , and the wave number @xmath59 is @xmath60 and @xmath61 is the experimental energy of the decaying state relative to the two - body breakup threshold .
as outlined in the previous section , we calculate @xmath2be + @xmath3 cluster states , using a previously published prescription for the potential @xcite , without adjusting any of the parameters @xmath29 , @xmath62 and @xmath63 listed in eq.(3 ) .
we identify the @xmath0 state with the hoyle state , check that this choice produces a good account of the available data , and proceed to calculate similar states of higher angular momenta .
each band of states is labelled by the choice of @xmath5 .
the maximum possible value of @xmath14 in this scheme is equal to @xmath35 itself ( corresponding to the nodeless wave function ) .
we determine the potential radius @xmath31 of the potential by fitting so as to reproduce the experimental energy of the @xmath0 hoyle state exactly .
the resulting values for @xmath31 are listed in table i. it is also useful to know the position and maximum height of the potential ( coulomb barrier ) since this gives an upper limit on the energy of a resonant state with the corresponding value of @xmath14 .
therefore we also list these values in table i for the three values of @xmath38 and 8 under consideration .
.[tab : table1 ] potential maxima for @xmath38 and 8 [ cols="^,^,^,^ " , ] figure 2 shows the energy dependence of @xmath64 and im(@xmath65 ) for the @xmath66 resonance near the c.m .
energy of about 2.22 mev .
this is both a typical case , and the state that we are most interested in ( because of recent experimental results @xcite which have at last found a rotational state of the hoyle band ) .
the best description of this state within our model is achieved using @xmath6 .
for @xmath37 there is no resonant state at all , and for @xmath67 the excitation energy is too low and the width too narrow .
we are not unduly concerned that this value of @xmath35 does not tally nicely " with oscillator shell model considerations , which might suggest g=@xmath68 since most of the low - lying intruder states in this region are described as 4p-4h excitations in the shell model .
we are not using a harmonic oscillator potential and will generate rather different wave functions with exponential rather than gaussian tails , and so do not expect an identity of @xmath35 values between the two cases .
our calculations also predict an excited @xmath39 state at 13.71 mev .
we predict a width of 1.24 mev for it , which would make it hard to detect in experiments .
we note however that this excitation energy places the state above the barrier maximum for an @xmath69 state in a @xmath6 band , indicating that the result is not completely reliable , although still indicative of a possible @xmath39 state in this region . in this context
it is interesting to note that freer et al.@xcite have recently found evidence for a new state in @xmath1c at @xmath70 mev with a width of @xmath71 mev in their studies of the @xmath1c(@xmath72he , @xmath72he+@xmath72he+@xmath72he ) @xmath72he and @xmath73be(@xmath72he , @xmath72he+@xmath72he+@xmath72he)n reactions .
analysis of the angular distributions suggests that the state might have j@xmath74 .
as such , its properties are in line with our expectations for the third state in the hoyle band , and we await the outcome of further investigations with interest .
previous analyses of inelastic electron scattering from @xmath1c @xcite indicate that the hoyle state has an abnormally large charge radius .
we therefore calculate the values of @xmath75 implied by our model for @xmath38 and 8 .
we first need to calculate a mean square charge radius for @xmath2be .
we do this by using the mean square charge radius for an @xmath3-particle of 1.6757 fm @xcite and the mean square separation of the two @xmath3-particles calculated by our bound state code and confirmed by gamow using the potential of ref.@xcite in the formula @xmath76 this yields a mean square charge radius for @xmath2be of 10.634 @xmath77 ( with a square root of 3.261 fm ) .
this , in turn , serves as input for the mean square charge radius of the hoyle state in the formula @xmath78 our @xmath6 value of 3.78 fm is somewhat below the value of 3.87 fm deduced by funaki et al .
@xcite but significantly larger than the values deduced from earlier gcm calculations of 3.50 fm @xcite and 3.47 fm @xcite obtained from full 3@xmath3 calculations using volkov effective two - nucleon forces .
it has long been suspected that the hoyle state in @xmath1c might have rotational excitations built upon it so that a hoyle band could be present in the @xmath1c spectrum .
recent experimental work has located a @xmath4 state a little below 10 mev with a width of about 600 kev which is a strong candidate for such a structure . this has motivated us to apply a local potential @xmath2be-@xmath3 cluster model , with previously published potential prescription , to the system to see if we can reproduce the existing data and predict properties of other similar states .
the calculation is numerically delicate but obtaining closely similar excitation energies from three different methods give us a good degree of confidence in our results .
we find that with a global quantum number of @xmath6 we are able to give a good account of the width and root mean square charge radius of the hoyle state itself and a reasonable description of the excitation energy and width of the proposed excited @xmath4 member of the putative hoyle band .
we note that our calculated @xmath4 state is very close to the top of the coulomb barrier .
we also predict a rather wide @xmath39 state of the @xmath6 band at an excitation energy of roughly 14 mev , and find that the band certainly terminates here ( if not already with the @xmath4 state ) .
however , our calculation is not completely reliable for this state , and there may be a case to be made that the known @xmath39 state at 14.083 mev @xcite should be assigned to the hoyle band .
perhaps there are even two @xmath39 states there .
this is an interesting conundrum which further experimental investigation might be able to resolve .
|
we describe the excited @xmath0 state of @xmath1c at 7.654 mev , often called the hoyle state , in terms of a local potential @xmath2be+@xmath3 cluster model .
we use a previously published prescription for the cluster - core potential to solve the schrdinger equation to obtain wave functions for this state , and also for higher angular momentum states of the same system .
we calculate energies , widths and charge radii for the resulting band of states , with particular emphasis on the recently discovered @xmath4 state .
we examine various choices of the global quantum number @xmath5 for the cluster - core relative motion , and find that @xmath6 leads to the most coherent description of the properties of the states and is consistent with recent experimental data on the @xmath7 state .
|
in the age of cloud computing , cloud servers with adequate resources help their clients by performing huge amount of computation or by storing large amount of data ( say , in the order of terabytes ) on their behalf . in this
setting , a client only has to download the result of the computation or has to read ( or update ) the required portion of the outsourced data .
several storage service providers like amazon simple storage service ( s3 ) , dropbox , google drive and microsoft azure provide storage outsourcing facility to their clients ( data owners )
. however , a cloud storage server can be malicious and delete some ( less frequently accessed ) part of the client s data in order to save space .
secure cloud storage protocols ( two - party protocols between the client and the server ) provide a cryptographic solution to this problem by ensuring that the client s data are stored untampered in the cloud server . in a secure cloud storage scheme , a client can remotely check the integrity of her data file outsourced to the cloud server . a possible way to do this
is to divide the data file into blocks and attach an authenticator ( or tag ) to each of these blocks before the initial outsourcing .
when the client wants to check the integrity of her data , she downloads all the blocks of the file along with their tags from the server and verifies them individually .
however , this process is highly inefficient due to the large communication bandwidth required between the client and the cloud server . in order to resolve the issue mentioned above
, a notion called _ proofs - of - storage _ comes into play where the client can audit her data file stored in the server without accessing the whole file , and still , be able to detect an unwanted modification of the file done by the malicious server .
the concept of _ provable data possession _ ( pdp ) is introduced by ateniese et al .
@xcite where the client computes an authentication tag for each block of her data file and uploads the file along with these authentication tags as stated earlier .
later , the client can execute an _ audit _ protocol and verify the integrity of the data file by checking only a predefined number of randomly sampled blocks of the file .
although efficient pdp schemes @xcite are available in the literature , they only provide the guarantee of retrievability of _ almost all _ blocks of the data file .
we briefly mention some situations where this guarantee is not sufficient .
the data file may contain some sensitive information ( e.g. , accounting information ) any part of which the client does not want to lose . on the other hand , a corruption in the compression table of an outsourced compressed file might make the whole file unavailable . in a searchable symmetric encryption scheme @xcite ,
the client encrypts a database using a symmetric encryption scheme to form an index ( metadata for that database ) @xcite and outsources the encrypted documents along with the index to the server .
the size of this index is very small compared to the encrypted database itself .
however , loss of the index completely defeats the purpose of the searchable encryption scheme . in such circumstances
, the client wants a stronger notion than pdp which would guarantee that the server has stored the _
entire _ file properly and the client can retrieve _ all _ of her data blocks at any point of time . to address the issue mentioned above , juels and
kaliski @xcite introduce _ proofs of retrievability _
( por ) where the data file outsourced to the server can be retrieved in its entirety by the client .
the underlying idea @xcite of a por scheme is to encode the original data file with an error - correcting code , authenticate the blocks of the encoded file with tags and upload them on the storage server . as in pdp schemes
, the client can execute an audit protocol to check the integrity of the outsourced data file .
the use of error - correcting codes ensures that _ all _ data blocks of the file are retrievable .
depending on the nature of the outsourced data , por schemes are classified as : por schemes for _ static _ data ( static por ) and _ dynamic _ data ( dynamic por ) .
for static data , the client can not change her data after the initial outsourcing ( suitable mostly for backup or archival data ) .
dynamic data are more generic in that the client can modify her data as often as needed .
the por schemes are _ publicly verifiable _ if audits can be performed by any third party auditor ( tpa ) with the knowledge of public parameters only ; they are _ privately verifiable _ if only the client ( data owner ) with some secret information can perform audits .
designing an _ efficient _ and publicly verifiable dynamic por scheme is an important research problem due to its practical applications .
there are various efficiency parameters where the performance of a dynamic por scheme might be improved .
some of them include the communication bandwidth required to read or write a data block ( or to perform an audit ) , the client s storage , and the computational cost at the client s ( or the server s ) end . on the other hand
, the client often prefers to delegate the auditing task to a third party auditor ( tpa ) who performs audits on the client s data and informs the client in case of any anomaly detected in the server s behavior . to the best of our knowledge , there is only one practical ) @xcite , and the current @xmath0 candidates are impractical . ]
dynamic por scheme with public verifiability that is proposed by shi et al .
@xcite . in this work ,
we provide a construction of a publicly verifiable dynamic por scheme that is more efficient ( in terms of write and audit costs ) than the publicly verifiable scheme proposed by shi et al .
moreover , the public parameters used in the latter scheme need to be changed for _ each _ write operation performed on the client s data which is clearly an overhead as , in that case , these public parameters are to be validated and certified for every write . *
our contribution*we summarize our contributions in this paper as follows .
* we construct a _ dynamic _ proofs - of - retrievability ( por ) scheme where the client outsources her data file to the cloud server and she can update the content of the file later . * our dynamic por scheme offers _ public verifiability _ , that is , the client can delegate the auditing task to a third party auditor who performs audits on the client s behalf . * we show that our scheme is secure in the sense that the client gets an assurance that her data file stored by the server is authentic and up - to - date , and _ all _ the data blocks can be retrieved by the client as often as needed . *
we analyze the performance of our scheme and compare it with other existing dynamic por schemes ( having private or public verifiability ) . *
our publicly verifiable dynamic por scheme enjoys more efficiency ( in terms of communication bandwidths required for a write and an audit ) than the `` state - of - the - art '' publicly verifiable dynamic por scheme @xcite .
moreover , unlike the latter scheme , there is no need to validate or certify the public parameters in our scheme for every write operation as they are fixed since the initial setup phase .
the rest of the paper is organized as follows .
section [ prelims ] describes the preliminaries and background related to our work . in section [ scs ] ,
we survey the existing literature on secure cloud storage schemes .
section [ scheme ] provides a detailed construction of our publicly verifiable dynamic por scheme .
we analyze the security of our scheme in section [ security ] .
finally , in section [ performance_ana ] , we discuss the performance of our scheme and compare our scheme with other existing dynamic por schemes based on different parameters ( shown in table [ tab : comparison_por ] ) .
we also show that our scheme is more efficient than the publicly verifiable dynamic por scheme proposed by shi et al . @xcite . in the concluding section [ sec : conclusion ]
, we summarize the work done in this paper .
we take @xmath1 to be the security parameter .
an algorithm denoted by @xmath2 is a probabilistic polynomial - time algorithm when its running time is polynomial in @xmath1 and its output @xmath3 is a random variable which depends on the internal coin tosses of @xmath4 .
if @xmath4 is given access to an oracle @xmath5 , we denote @xmath4 by @xmath6 also .
an element @xmath7 chosen uniformly at random from a set @xmath8 is denoted as @xmath9 .
a function @xmath10 is called negligible in @xmath1 if for all positive integers @xmath11 and for all sufficiently large @xmath1 , we have @xmath12 . a @xmath13-erasure code is an error - correcting code @xcite that comprises an encoding algorithm enc : @xmath14 ( encodes a message consisting of @xmath15 symbols into a longer codeword consisting of @xmath16 symbols ) and a decoding algorithm dec : @xmath17 ( decodes a codeword to a message ) , where @xmath18 is a finite alphabet and @xmath19 is the minimum distance ( hamming distance between any two codewords is at least @xmath19 ) of the code .
the quantity @xmath20 is called the rate of the code .
a @xmath13-erasure code can tolerate up to @xmath21 erasures .
if @xmath22 , we call the code a maximum distance separable ( mds ) code . for an mds code , the original message can be reconstructed from any @xmath15 out of @xmath16 symbols of the codeword .
reed - solomon codes @xcite and their extensions are examples of non - trivial mds codes .
a merkle hash tree @xcite is a binary tree where each leaf - node stores a data item .
the label of each leaf - node is the data item stored in the node itself .
a collision - resistant hash function @xmath23 is used to label the intermediate nodes of the tree .
each of the outputs of @xmath23 on different inputs is a binary string of length @xmath24 .
the label of a intermediate node @xmath25 is the output of @xmath23 computed on the labels of the children nodes of @xmath25 .
a merkle hash tree is used as a standard tool for efficient memory - checking .
child node b + @xmath26 child node d + @xmath27 child node h + @xmath28 child node i + @xmath29 child node e + @xmath30 child node j + @xmath31 child node k + @xmath32 child node c + @xmath33 child node f + @xmath34 child node l + @xmath35 child node m + @xmath36 child node g + @xmath37 child node
n + @xmath38 child node o + @xmath39 ; fig .
[ fig : mht ] shows a merkle hash tree containing the data items @xmath40 stored at the leaf - nodes .
consequently , the labels of the intermediate nodes are computed using the hash function @xmath23 .
the hash value of the node @xmath41 is the _ root - digest_. the proof showing that a data item @xmath19 is present in the tree consists of the data item @xmath19 and the labels of the nodes along the _ associated path _ ( the sequence of siblings of the node containing the data item @xmath19 ) .
for example , a proof showing that @xmath31 is present in the tree consists of @xmath42 , where @xmath43 and @xmath44 are the labels of the nodes @xmath45 and @xmath46 , respectively .
given such a proof , a verifier computes the hash value of the root .
the verifier outputs ` accept ` if the computed hash value matches with the root - digest ; it outputs ` reject ` , otherwise .
the size of a proof is logarithmic in the number of data items stored in the leaf - nodes of the tree . due to the collision - resistance property of @xmath23 , it is infeasible ( except with some probability negligible in the security parameter @xmath1 ) to add or modify a data item in the merkle hash tree without changing its root - digest .
we define a digital signature scheme as proposed by goldwasser et al .
a digital signature scheme consists of the following polynomial - time algorithms : a key generation algorithm keygen , a signing algorithm sign and a verification algorithm verify .
keygen takes as input the security parameter @xmath1 and outputs a pair of keys @xmath47 , where @xmath48 is the secret key and @xmath49 is the corresponding public verification key .
algorithm sign takes a message @xmath50 from the message space @xmath51 and the secret key @xmath48 as input and outputs a signature @xmath52 .
algorithm verify takes as input the public key @xmath49 , a message @xmath50 and a signature @xmath52 , and outputs ` accept ` or ` reject ` depending upon whether the signature is valid or not .
any of these algorithms can be probabilistic in nature .
the correctness and security ( existential unforgeability under adaptive chosen message attacks @xcite ) of a digital signature scheme are described as follows . 1 . _
correctness_:algorithm verify always accepts a signature generated by an honest signer , that is , @xmath53=1 .
\end{aligned}\ ] ] 2 . _
security_:let sign@xmath54 be the signing oracle and @xmath4 be any probabilistic polynomial - time adversary with an oracle access to sign@xmath54 .
the adversary @xmath4 adaptively makes polynomial number of sign queries to sign@xmath54 for different messages and gets back the signatures on those messages .
the signature scheme is secure if @xmath4 can not produce , except with some probability negligible in @xmath1 , a valid signature on a message not queried previously , that is , for any probabilistic polynomial - time adversary @xmath55 , the following probability is negligible in @xmath1 , where @xmath56 is the set of sign queries made by @xmath4 to @xmath57 .
the discrete log problem over a multiplicative group @xmath58 of prime order @xmath59 and generated by @xmath60 is defined as follows .
given @xmath61 , the discrete log problem over @xmath62 is to compute @xmath63 such that @xmath64 .
the discrete log assumption says that , for any probabilistic polynomial - time adversary @xmath2 , the following probability _
x_q[x(y ) : y = g^x ] is negligible in @xmath1 , where the probability is taken over the internal coin tosses of @xmath4 and the random choice of @xmath65 .
we define a proofs - of - retrievability scheme for _ dynamic _ data as follows @xcite . a dynamic por scheme consists of the following protocols between two stateful parties : a client ( data owner ) and a server . * _ init(@xmath66 ) : _ this protocol associates a random file - identifier _ `
fid ` _ to the data file @xmath67 consisting of @xmath68 data blocks each of @xmath69 bits , and it outputs the client state @xmath70 and another file @xmath71 to be stored by the server . * _ read(@xmath72 ) : _ this protocol outputs the data block at the @xmath73-th location of the current state of the file or _ ` abort`_. * _ write(@xmath74 ) : _ this protocol inserts the block @xmath75 after the @xmath73-th block of the file or sets @xmath73-th block of the file to @xmath75 or deletes the @xmath73-th block of the file ( @xmath75 is _ ` null ` _ in this case ) .
it outputs updated @xmath76 or _ ` abort`_. * _ audit(@xmath77 ) : _ this protocol checks memory contents of the current state of the data file and outputs 1 or 0 .
a dynamic por scheme is _ privately verifiable _ if only the client with some secret information can perform an audit , that is , @xmath70 is secret .
otherwise , it is _
publicly verifiable_. for a publicly verifiable dynamic por scheme , a third party auditor ( tpa ) can audit the client s data on behalf of the client who delegates her auditing task to the tpa . in this case
, we use the term `` verifier '' to denote an auditor who can be the tpa or the client herself .
security of a dynamic por scheme is described in section [ security_def ] .
ateniese et al .
@xcite introduce the notion of _ provable data possession _ ( pdp ) . in a pdp scheme , the client computes an authentication tag ( e.g. , message authentication code @xcite ) for each block of her data file and uploads the file along with the authentication tags . during an audit protocol
, the client samples a predefined number of random block - indices and sends them to the server ( _ challenge _ phase ) .
the cardinality of the challenge set is typically taken to be @xmath24 , where @xmath1 is the security parameter .
depending upon the challenge , the server does some computations over the stored data and sends a proof to the client ( _ response _ phase ) .
finally , the client checks the integrity of her data based on this proof ( _ verification _ phase ) .
_ almost all _ data blocks can be retrieved from a ( possibly malicious ) server passing an audit with a non - negligible probability .
other pdp schemes include @xcite . juels and kaliski @xcite introduce _ proofs of retrievability _
( por ) for static data ( naor and rothblum @xcite give a similar idea for sublinear authenticators ) . according to shacham and waters @xcite , the retrievability guarantee for _ all _ data blocks of the outsourced
file can be achieved by encoding the original file with an erasure code ( see section [ erasure_code ] ) before authenticating ( and uploading ) the blocks of the encoded file . due to
the redundancy added to the data blocks , the server has to delete or modify a considerable number of blocks to actually delete or modify a data block which makes it difficult for the server to pass an audit . following the work by juels and kaliski , several por schemes have been proposed for static data ( _ static por _ ) and dynamic data ( _ dynamic por _ ) .
shacham and waters @xcite propose two por schemes for static data ( one with private verifiability and another with public verifiability ) where the response from the server is short .
bowers et al .
@xcite propose a theoretical framework for designing por schemes and provide a prototype implementation of a variant of @xcite . in another work , bowers et al .
@xcite introduce hail ( high - availability and integrity layer ) , a distributed por setting where the client s data are disseminated across multiple servers .
dodis et al .
@xcite introduce a notion called `` por codes '' and show how por schemes can be instantiated based on these por codes .
they explore the connection between por codes and the notion of hardness amplification @xcite .
xu and chang @xcite improve the privately verifiable scheme of @xcite by making the communication bandwidth required for an audit to be @xmath24 .
armknecht et al .
@xcite propose a por scheme where any entity among the client ( data owner ) , the auditor and the cloud server can be malicious , and any two of them can collude as well .
the auditor performs two audits : one for the auditor itself and another on behalf of the client .
the challenge sets are generated using a public randomized algorithm derived from the hash value of the latest block added to the bitcoin block chain @xcite .
a few dynamic por schemes are there in the literature .
stefanov et al .
@xcite propose an authenticated file system called `` iris '' that is highly scalable and resilient against a malicious cloud server .
cash et al .
@xcite encode a small number of data blocks _ locally _ and hide the access pattern of the related ( belonging to the same codeword ) blocks from the server using oblivious ram ( oram ) @xcite .
due to the use of expensive oram primitives , this scheme is inefficient .
shi et al .
@xcite propose two practical dynamic por schemes which reduce the cost of computation as well as the communication bandwidth required to execute the protocols involved .
chandran et al .
@xcite introduce the notion of `` locally updatable and locally decodable codes '' and propose an efficient dynamic por scheme by applying the techniques used in the construction of such a code . the por scheme by guan et al .
@xcite exploits the privately verifiable scheme of @xcite and gives a publicly verifiable scheme with the help of indistinguishability obfuscation ( @xmath0 ) @xcite .
in this section , we describe our publicly verifiable dynamic por scheme with efficient writes and audits . like the existing dynamic por schemes @xcite ,
our construction also rely on the hierarchical structure provided by the oblivious ram @xcite .
specifically , we follow a storage structure similar to the one proposed by shi et al . @xcite . however , our construction is more efficient than their scheme in terms of the cost of a write operation and the cost of an audit .
our construction is based on _ collision - resistant homomorphic hashing _
technique @xcite along with a digital signature scheme .
our scheme relies on a storage structure similar to that proposed by shi et al .
let the client ( data owner ) associate a random file - identifier ` fid ` of @xmath1 bit - size to the data file she wants to outsource to the cloud server .
we assume that the data file is divided into blocks of size @xmath69 bits , and read ( and write ) operations are performed on these blocks .
the value of @xmath69 is taken to be @xmath78 for a large prime @xmath79 .
the way this prime @xmath79 is selected is discussed in section [ buff_h ] . for static data , a standard way to guarantee retrievability of the file is to encode the file using an erasure code @xcite .
the main drawback of using erasure codes in _ dynamic _ por is that an update in a single block in a codeword ( say , c ) is followed by updates on other @xmath80 blocks in c , where @xmath68 is the number of blocks being encoded to form c. the underlying idea to overcome this drawback is not to update the encoded copy ( c ) for every write operation ( insertion , deletion or modification ) .
instead , it is updated ( or rebuilt ) only when sufficient updates are done on the data file .
thus , the amortized cost for writes is reduced dramatically . however ,
this encoded copy stores stale data between two such rebuilds . therefore ,
a hierarchical log structure is maintained which temporarily stores values for the intermediate writes between two successive rebuilds of c. each level of this hierarchical log is also encoded using an erasure code .
we adopt the storage structure and code construction mentioned above in our scheme .
however , we use collision - resistant homomorphic hashing to construct another hierarchical storage ( discussed in section [ storage_tags ] ) in order to reduce the cost of a write and an audit for the client .
our scheme involves the following three data structures in order to store the data blocks of the client s file : * an _ unencoded _ buffer u containing all the up - to - date data blocks of the file ( that is , u is updated after every write operation is performed ) , * an _ encoded _ buffer c which is updated after every @xmath68 writes ( that is , c is rebuilt afresh by encoding the latest u after every @xmath68 write operations ) , and * an _ encoded _ hierarchical ( multiple levels of buffers ) log structure h which accommodates all intermediate writes between two successive rebuilds of c ( h is made empty after every @xmath68 writes ) .
we note that _ all _ of these data structures are stored on the cloud server .
the server also stores two additional data structures , @xmath81 and @xmath82 ( similar to h and c , respectively ) , for authentication tags described in section [ storage_tags ] .
the buffer u contains an up - to - date copy of the data file .
reads and writes are performed directly on the required locations of u. a merkle hash tree is maintained over the data blocks of u to check the authenticity of the read block ( see section [ mht ] for the description of a merkle hash tree ) . the merkle proof sent by the server along with the read block
is verified with respect to the up - to - date root - digest ( say , @xmath83 ) of the merkle hash tree .
one can also use other authenticated data structures like rank - based authenticated skip lists @xcite instead of a merkle hash tree .
let @xmath68 be the number of blocks the client outsources to the cloud server initially .
so the height of the merkle tree built on u is @xmath84 .
read and write operations on the buffer u are described in details in section [ operations ] .
a hierarchical log structure h is maintained that consists of @xmath85 levels @xmath86 , where @xmath87 .
the log structure h stores the intermediate writes temporarily . for each @xmath88 ,
the @xmath89-th level h@xmath90 consists of an encoded copy of @xmath91 data blocks using a @xmath92-erasure code , where @xmath93 and @xmath94 contain @xmath91 blocks each .
the original data blocks encoded in h@xmath95 arrive at time @xmath96 , where @xmath97 is a multiple of @xmath91 .
we describe the encoding procedure as follows .
let @xmath79 be a large prime such that @xmath98 for some @xmath99 and the bit - size of a block @xmath100 , where @xmath101 .
let @xmath102 denote a generator of @xmath103 .
then , @xmath104 is a @xmath105-th primitive root of unity modulo @xmath79 .
when a block @xmath75 is written to h , it is inserted in the topmost level ( @xmath106 ) if h@xmath107 is empty .
that is , @xmath108 is set to @xmath75 .
in addition , @xmath109 is set to @xmath110 for the @xmath97-th ( @xmath111 ) write , where @xmath112 is the @xmath105-th primitive root of unity modulo @xmath79 .
here , @xmath113 is the bit reversal function , where @xmath114 is the value of the binary string obtained by reversing the binary representation of @xmath97 . if the top @xmath73 levels @xmath115 are already full , a _
rebuild _ is performed to accommodate all the blocks in these levels as well as the new block in @xmath116 ( and to make all the levels up to @xmath117 empty ) .
shi et al .
@xcite employ a fast incrementally constructible code based on fast fourier transform ( fft ) @xcite , we use similar code and parameters for the ease of comparison . ] .
[ fig : rebuildx ] describes the algorithm for rebuild of @xmath93 that in turn uses the algorithm ` mix ` shown in fig .
[ fig : mixx ] .
although the algorithm deals with @xmath93 , the same algorithm can be used for rebuilding @xmath94 if we replace the @xmath118 arrays by corresponding @xmath119 arrays and the incoming block @xmath75 by @xmath110 .
we refer @xcite for the form of the @xmath120 generator matrix @xmath121 for the @xmath89-th level code .
let @xmath122 be the vector containing @xmath91 data blocks most recently inserted in h ( after applying a permutation ) .
then , the output of the algorithm ` mix ` for h@xmath95 is the same as that when @xmath122 is multiplied by @xmath121 .
any @xmath123 submatrix of the generator matrix @xmath121 is full rank , and thus , the code achieves the maximum distance separable ( mds ) property . as a concrete example
, the rebuild of @xmath124 is demonstrated in fig .
[ fig : rebuildh ] .
the other part of h@xmath125 ( that is , @xmath126 ) is rebuilt in a similar fashion .
we observe that , by using this code , the rebuild cost of h@xmath95 is @xmath127 ( i.e. , linear in the length of h@xmath95 ) since the algorithm ` mix ` populates h@xmath95 by combining two arrays @xmath128 ( see fig . [
fig : mixx ] and fig .
[ fig : rebuildh ] ) .
the @xmath89-th level is rebuilt after @xmath91 writes .
therefore , the amortized cost for rebuilding is @xmath129 per write operation .
each rebuild of the buffer c ( discussed in section [ buff_c ] ) is followed by making all levels of h empty .
unlike the buffer u ( and h ) , no read or write operations are performed directly on the buffer c. after @xmath68 write operations , the buffer u is encoded using an erasure code to form a new copy of c , and the existing copy of c is replaced by this new one .
the rebuild of c can be done using the same fft - based code discussed in section [ buff_h ] which costs @xmath130 both in time and bandwidth . as c is rebuilt after every @xmath68 write operations , the cost incurred per write is @xmath129 .
we note that c contains stale data between two successive rebuilds .
however , the intermediate writes are accommodated in h with appropriate encoding .
thus , these blocks written between two successive rebuilds of c are also retrievable at any point of time .
in addition to the log structure h and the buffer c for data blocks , two similar structures @xmath81 and @xmath82 for authentication tags corresponding to the data blocks in h and c ( respectively ) are stored on the cloud server .
thus , the server stores u , h , c , @xmath81 and @xmath82 .
the benefits of storing @xmath81 and @xmath82 on the server are as follows .
let us assume that the authentication tags for data blocks have the following properties . 1 .
the size of a tag is much less than that of a block .
the tags are homomorphic in the sense that , given the tags of two blocks @xmath131 and @xmath132 , the tag on any linear combination of @xmath131 and @xmath132 can be generated .
we note that the fundamental operation for a write ( or rebuild ) on h and c is to encode data blocks , that is , to compute a linear combination of those data blocks ( see eqn . 1 and eqn . 2 in fig .
[ fig : mixx ] ) . due
the _ second _ property mentioned above , while the server itself performs write operations on h and c , the client ( data owner ) can perform similar operations on @xmath81 and @xmath82 . on the other hand , due to the _ first _ property
, the bandwidth required between the client and the server for a write operation decreases significantly as the client now has to download much smaller tags instead of large data blocks .
the cost of storage is less nowadays , whereas bandwidth is more expensive and often limited .
therefore , it is reasonable if we trade the storage off to reduce the communication bandwidth between the client and the server required for a write ( or rebuild ) .
indeed , the authentication tags ( described in the following section ) we use in our dynamic por scheme satisfy the following properties . *
the size of a tag is @xmath24 and is independent of the size of a data block ( @xmath1 is the security parameter ) , where @xmath133 . *
the homomorphic property is achieved by using a collision - resistant homomorphic hash function .
apart from efficient write operations , the cost of an audit in our _ publicly verifiable dynamic _ por scheme is comparable to that in the privately verifiable scheme of @xcite , and it is _ much less _ than that in the publicly verifiable scheme discussed in the same work . *
* setup**for the data file identified by ` fid ` , the client runs an algorithm setup(@xmath134 ) to set parameters for generating authentication tags .
the algorithm setup selects two random primes @xmath135 and @xmath59 such that @xmath136 , @xmath137 and @xmath138 .
now , it divides each block @xmath75 of the data file into segments of size @xmath139 bits each .
this ensures that each segment is less than @xmath59 and can therefore be represented as an element of @xmath140 .
thus , @xmath141 is the number of segments in a block , where a block is @xmath100 bits long . in this
setting , each block @xmath75 can be represented as a vector @xmath142\in \z_q^m$ ] .
let @xmath62 be a subgroup of @xmath143 having order @xmath59 .
that is , @xmath62 consists of the order @xmath59 elements in @xmath144 .
then , @xmath50 random elements @xmath145 are selected .
let @xmath146 be a digital signature scheme ( see section [ dig_sig ] ) where the algorithm sign takes messages in @xmath147 as input and outputs signatures of size @xmath24 bits each .
let the pair of signing and verification keys for @xmath148 be @xmath149 .
the setup algorithm outputs the primes @xmath135 and @xmath59 , the secret key @xmath150 , the public parameters @xmath151 , and the descriptions of @xmath62 and @xmath148 . * * format of a tag**the client computes the homomorphic hash @xcite on a block @xmath152\in \z_q^m$ ] as @xmath153 using the signature scheme @xmath148 , the client generates the final authentication tag for the block @xmath75 as @xmath154 where ` addr ` is the physical address the data block @xmath75 is written to at time @xmath97 and ` fid ` is the file - identifier of the data file the block @xmath75 belongs to . * * collision - resistance and homomorphic properties**as shown in @xcite , given that the discrete log assumption ( see section [ dislog ] ) holds in @xmath62 , it is computationally hard to find two data blocks @xmath131 and @xmath132 such that @xmath155 and @xmath156 ( _ collision - resistance _ property ) . on the other hand , given two data blocks @xmath157\in \z_q^m$ ] and @xmath158\in \z_q^m$ ] , any linear combination of @xmath131 and @xmath132 can be written as @xmath159\in \z_q^m$ ] .
therefore , @xmath160 can be computed as @xmath161 ( _ homomorphic _ property ) . * * size of a tag**the size of an authentication tag @xmath162 is the sum of @xmath163 ( that is , @xmath164 bits ) and the size of signature of @xmath148 . if we use the standard ecdsa ( elliptic curve digital signature algorithm ) @xcite as the signature scheme @xmath148 , then each signature is @xmath165 bits long bits @xcite .
however , the verification of a signature is more expensive due to computation of bilinear pairings @xcite . ] .
thus , @xmath166 is also @xmath24 bits . * * improvement in cost of tag computation**to compute the homomorphic hash @xmath160 on a block @xmath75 using eqn .
[ eqn : tag ] , it requires @xmath50 exponentiations and @xmath167 multiplications modulo @xmath135 .
we can reduce this computational complexity in the following way at the cost of the client storing @xmath50 elements of @xmath168 which is essentially equivalent to just one block .
the client chooses @xmath169 and @xmath170 .
the client sets @xmath171 for each @xmath172 $ ] .
the client includes the vector @xmath173 $ ] and @xmath60 in her secret key @xmath174 and makes @xmath175 public as before .
now , the homomorphic hash @xmath160 on a block @xmath75 is computed as @xmath176 which requires only _ one _ exponentiation modulo @xmath135 along with @xmath50 multiplications and @xmath167 additions modulo @xmath59 .
this is a huge improvement in the cost for computing an authentication tag . on the other hand ,
the storage overhead at the client s side is @xmath177 which is same as the size of a single block @xmath75 .
considering the fact that the client outsources millions of blocks to the cloud server , this amount of client storage is reasonable for all practical purposes .
the storage structures for @xmath81 and @xmath82 are exactly the same as those for h and c , respectively , except that the authentication tags ( instead of data blocks ) are stored in @xmath81 and @xmath82 ( see section [ storage_blocks ] for structures of h and c ) .
there are three types of operations involved in a dynamic por scheme .
the client can read , write and audit her data stored on the cloud server .
the read and write operations are _ authenticated _ in that the client can verify the authenticity of these operations .
we note that though the client herself performs reads and writes on her data , she can delegate the auditing task to a third party auditor ( tpa ) for a publicly verifiable scheme .
as our scheme is _ publicly verifiable _ , we use the term _ verifier _ to denote an auditor who can be a tpa or the client herself .
[ fig : flow ] gives an overview of the communication flow between the client and the server during these operations .
reads are performed directly from the unencoded buffer u. the authenticity and freshness of the block read can be guaranteed by using a merkle hash tree @xcite ( or a similar data structure like rank - based authenticated skip list @xcite ) over the blocks of u. that is , the blocks of u constitute the leaf - nodes of the merkle hash tree ( see section [ mht ] for a brief description of a merkle hash tree ) . the server sends the merkle proof @xmath178 containing the requested block and the labels of the nodes along the associated path of the tree to the client .
the client maintains the up - to - date value of the root - digest of the merkle hash tree ( or the label of the root node of the rank - based authenticated skip list ) , and she verifies the proof @xmath178 with respect to this root - digest .
we note that the size of the root - digest of the merkle hash tree is @xmath24 bits .
let ` updtype ` denote the type of a write operation which can be insertion of a new block after the @xmath73-th block , deletion of the @xmath73-th block or modification of the @xmath73-th block .
a write operation affects the buffers in the following way . *
* * write on u**as the buffer u is unencoded , a write operation on u can be performed in a similar way as done on the data blocks in a dynamic provable data possession ( pdp ) scheme @xcite .
+ let @xmath83 be the root - digest of the current merkle hash tree which is stored at the client s side .
the client performs an _ authenticated _ read on the @xmath73-th data block of u ( as described above ) . if the associated merkle proof @xmath178 does not match with @xmath83 , the client aborts
otherwise , she computes , from @xmath178 , the value that would be the new root - digest ( say , @xmath179 ) if the write operation is correctly performed on u stored at the server .
the client stores @xmath179 temporarily at her end and asks the server to perform the write on u. the server performs this write operation on u , computes the root - digest @xmath180 of the merkle hash tree and sends @xmath180 to the client .
the client verifies whether @xmath181 if they are not equal , the client aborts
. otherwise , the client sets @xmath182 at her end . *
* * write on h and @xmath81**we assume that deletion of a block in u corresponds to insertion of a block ( with ` null ` content ) in the hierarchical log h. therefore , for a write of any ` updtype ` ( i.e. , insertion , deletion or modification ) , only insertions take place in h. the way a ( possibly encoded ) block @xmath75 is inserted in h is discussed in details in section [ buff_h ] .
the cloud server itself performs this operation on h. + an insertion in @xmath81 is performed by the client herself as this procedure requires the knowledge of secret information held by the client .
the client computes the authentication tag on the ( possibly encoded ) block and insert it in @xmath81 .
the underlying basic operation of the rebuild phase of h is to compute a linear combination @xmath75 ( e.g. , @xmath183 of two blocks @xmath131 and @xmath132 ( see eqn . 1 and 2 in fig .
[ fig : mixx ] ) .
similarly , the corresponding operation for the rebuild of @xmath81 is to compute @xmath162 given @xmath184 and @xmath185 . for @xmath186 ,
the client first downloads @xmath187 and verifies the signature on @xmath188 by checking whether @xmath189 where @xmath49 is the verification key for the signature scheme @xmath148 .
we note that ` addr`@xmath190 ( or ` addr`@xmath191 ) is the physical address of the block @xmath131 ( or @xmath132 ) written at time @xmath192 ( or @xmath193 ) , and ` fid ` is the file - identifier of the data file the block @xmath75 belongs to . for any block in h and c ,
the time when the block was written most recently can be easily computed from the current time itself .
if the verification passes , the client computes @xmath194 and @xmath162 subsequently .
this requires two exponentiations and one multiplication modulo @xmath135 along with one sign and two verify operations . * * * write on c and @xmath82**as mentioned in section [ buff_c ] , c ( @xmath82 for authentication tags ) is rebuilt after every @xmath68 writes .
the server performs a rebuild on c , whereas a rebuild on @xmath82 is performed by the client .
basic operations involved in rebuilds of c and @xmath82 are the same as those for rebuilds of h and @xmath81 , respectively , and thus are omitted here . in the _ challenge _ phase , the verifier chooses @xmath195 random locations @xmath196 from different levels ( @xmath24 locations from each level ) of h and c. then , she sends a challenge set @xmath197 to the cloud server , where @xmath198 are random coefficients . in the _ response _ phase
, the server sends to the verifier a proof containing @xmath199 and @xmath200 . upon receiving the proof from the server , the verifier verifies each of the signatures on @xmath201 .
then , she computes @xmath202 and @xmath203 using eqn .
[ eqn : tag ] . finally , the verifier checks whether @xmath204 and outputs 0 if any of the verifications fails ; she outputs 1 , otherwise .
we define security of a dynamic por scheme @xcite and show that our scheme described in section [ scheme ] is secure according to this definition .
we also show that the server can not pass an audit without storing _ all _ data blocks properly , except with some probability negligible in @xmath1 .
a secure dynamic por scheme offers the following guarantees . * * * authenticity and freshness**the authenticity property requires that the cloud server can not produce valid proofs during authenticated reads , writes and audits without storing the corresponding blocks and their respective authentication information untampered , except with a probability negligible in the security parameter @xmath1 .
+ for dynamic data , the client can modify an existing data block .
however , a malicious cloud server may discard this change and keep an old copy of the block .
as the old copy of the block and its corresponding tag constitute a valid pair , the client has no way to detect if the cloud server is storing the _ fresh _ ( latest ) copy .
thus , the client must be convinced that the server has stored the up - to - date blocks . * * * retrievability**we first describe the following security game between the challenger and the adversary . * * the adversary selects a file @xmath67 associated with a file - identifier ` fid ` to store .
the challenger processes the file to form another file @xmath71 and returns @xmath71 to the adversary .
the challenger stores only some metadata for verification purpose . * * the adversary adaptively chooses a sequence of operations defined by @xmath205 ( @xmath206 is polynomial in the security parameter @xmath1 ) , where @xmath207 is a read , a write or an audit .
the challenger executes these operations on the file stored by the adversary . for each operation
, the challenger verifies the response sent by the adversary and updates the metadata at her end only if the response passes the verification . *
* let @xmath208 be the final state of the file after @xmath206 operations .
the challenger has the latest metadata for the file @xmath208 .
now , she executes an audit protocol with the adversary .
the challenger sends a random challenge set @xmath209 to the adversary , and the adversary returns a cryptographic proof to the challenger .
the adversary wins the game if the verification algorithm accepts the proof .
+ the retrievability of data requires that , given a probabilistic polynomial - time adversary @xmath4 that can win the security game described above with some non - negligible probability , there exists a polynomial - time extractor algorithm @xmath210 that can extract _ all _ the blocks ( except with negligible probability ) by challenging @xmath4 for a polynomial ( in @xmath1 ) number of times and verifying the responses sent by @xmath4 .
the algorithm @xmath210 has _ black - box rewinding access _ to @xmath4 .
our scheme satisfies the following properties required for security . *
* authenticity and freshness * the authenticity and freshness of reads and writes on u are guaranteed by the use of a merkle hash tree on the blocks of u. every read or write operation is associated with a merkle proof ( corresponding to the up - to - date root - digest ) which is hard to forge due to the collision - resistant hash function @xmath23 used in the construction of the merkle hash tree ( see section [ mht ] ) . + as discussed in section [ tag_gen ] , due to the collision - resistance property of the authentication tags , it is computationally hard to find two data blocks @xmath131 and @xmath132 such that @xmath155 and @xmath156 . additionally , an authentication tag for a block @xmath75 is of the form @xmath211 , where the signing algorithm sign uses the secret key @xmath48 of the client and @xmath97 is the last write - time ( computable from the current time ) of the block @xmath75 .
the authenticity and freshness of h , @xmath81 , c and @xmath82 ( checked during an audit ) directly follow from these facts combined . * * * retrievability**given that our scheme satisfies the _ authenticity _ and _ freshness _ properties mentioned above , it is not hard to see that there exists a polynomial - time extractor algorithm @xmath210 , for each adversary @xmath4 that wins the security game described in section [ security_def ] with some non - negligible probability , that can extract at least @xmath212-fraction of the data blocks ( for some @xmath212 ) with the help of gaussian elimination @xcite .
use of a @xmath212-rate erasure code ensures retrievability of all the blocks of the data file . in our case , @xmath213 .
as we mention in section [ operations ] , all the levels of h and the buffer c are audited with @xmath24 random locations each . to delete a block in any of these levels
, the server has to actually delete half of the blocks in that level .
this is true due to the use of a @xmath92-erasure code for each level @xmath214 .
thus , if the server has corrupted half of the blocks in any level , then the server passes an audit with probability @xmath215 which is negligible in the security parameter @xmath1 .
we analyze the performance of the following types of operations ( described in section [ operations ] ) involved in our publicly verifiable dynamic por scheme . * * * read**for an authenticated read on the data block @xmath75 present in u , the server sends the corresponding merkle proof @xmath178 which consists of the block @xmath75 , the data block in the sibling leaf - node of @xmath75 and the hash values along the associated path of the merkle hash tree ( see section [ mht ] ) .
thus , a read operation takes @xmath216 communication bandwidth between the client and the server .
+ to reduce this cost , the client can generate authentication tags on the data blocks of u ( as discussed in section [ tag_gen ] ) and construct a merkle tree over these tags instead of the data blocks . in this setting , @xmath178 consists of @xmath162 , the authentication tag in its sibling leaf - node and the hash values along the associated path .
this reduces the communication bandwidth between the client and the server for a read to @xmath217 . * * * write**a write operation incurs the following costs . *
* _ write on u_:a write operation on u involves an authenticated read operation followed by the verification of eqn .
[ eqn : mht ] .
thus , each write operation requires @xmath217 bandwidth between the client and the server ( for communicating @xmath178 and @xmath180 ) . *
* _ write on h and @xmath81_:the cost of a write on h is @xmath129 ( see section [ buff_h ] ) . similarly ,
the cost of a write on @xmath81 is @xmath218 as the blocks are replaced by their authentication tags in @xmath81 and the size of a tag is @xmath24 bits . * * _ write on c and @xmath82_:c ( or @xmath82 ) is rebuilt after every @xmath68 writes .
as mentioned in section [ buff_c ] , a write operation on c costs @xmath129 both in time and bandwidth .
similarly , the cost of a write on @xmath82 is @xmath218 .
* * * audit**during an audit , for a challenge set @xmath209 containing @xmath195 random locations @xmath196 and random coefficients @xmath219 , the server computes and sends to the verifier a proof containing @xmath199 and @xmath200 .
thus , the bandwidth required for an audit is given by @xmath220 . * * comparison among dynamic por schemes**we compare our scheme with other existing dynamic por schemes which is summarized in table [ tab : comparison_por ] .
the comparison is based on the asymptotic complexity for different parameters .
some of the figures mentioned in table [ tab : comparison_por ] are taken from @xcite .
table [ tab : parameters ] mentions typical values of some parameters used in our scheme @xcite .
.comparison among dynamic por schemes based on different parameters ( asymptotic complexity ) [ cols="^,^,^,^,^,^,^ " , ] additionally , one drawback of the publicly verifiable scheme proposed by shi et al .
@xcite is due to the fact that one or more merkle hash trees ( _ separate _ from the merkle hash tree maintained for u ) are maintained to ensure the integrity of the blocks in the _ hierarchical log _ h ( one for the entire log or one for each of its levels ) . to enable a third party auditor ( tpa ) to audit the data blocks residing at different levels of this log , the root - digests of these trees need to be made public
however , some of these root - digests are changed as often as new data blocks are inserted in the hierarchical log structure , thus resulting in a change in the public parameters for _ each _ write .
this incurs an additional ( non - trivial ) overhead for validation and certification of the public parameters for every write operation . on the other hand ,
the public parameters in our publicly verifiable dynamic por scheme are fixed throughout the execution of the protocols involved .
in this work , we have proposed a dynamic por scheme where the client can update her data file after the initial outsourcing of the file to the cloud server and retrieve all of her data at any point of time .
our scheme is publicly verifiable , that is , anyone having the knowledge of the public parameters of the scheme can perform an audit on the client s behalf .
our scheme offers security guarantees of a dynamic por scheme .
our publicly verifiable dynamic por scheme is more efficient ( in terms of the cost of a write or an audit ) than the only known construction of a practical and publicly verifiable dynamic por scheme @xcite .
ateniese , g. , burns , r.c . ,
curtmola , r. , herring , j. , kissner , l. , peterson , z.n.j . , song , d.x . : provable data possession at untrusted stores . in : acm conference on computer and communications security , ccs 2007 .
( 2007 ) 598609 ateniese , g. , pietro , r.d . , mancini , l.v .
, tsudik , g. : scalable and efficient provable data possession . in : international conference on security and privacy in communication networks ,
securecomm 2008 .
( 2008 ) 9 curtmola , r. , garay , j.a . , kamara , s. , ostrovsky , r. : searchable symmetric encryption : improved definitions and efficient constructions . in : acm conference on computer and communications security , ccs 2006 .
( 2006 ) 7988 garg , s. , gentry , c. , halevi , s. , raykova , m. , sahai , a. , waters , b. : candidate indistinguishability obfuscation and functional encryption for all circuits . in : ieee symposium on foundations of computer science , focs 2013 .
( 2013 ) 4049 guan , c. , ren , k. , zhang , f. , kerschbaum , f. , yu , j. : symmetric - key based proofs of retrievability supporting public verification . in : european symposium on research in computer security - esorics 2015 , part i. ( 2015 ) 203223 impagliazzo , r. , jaiswal , r. , kabanets , v. : approximately list - decoding direct product codes and uniform hardness amplification . in : ieee symposium on foundations of computer science - focs 2006 .
( 2006 ) 187196 wang , q. , wang , c. , ren , k. , lou , w. , li , j. : enabling public auditability and data dynamics for storage security in cloud computing .
transactions on parallel and distributed systems * 22*(5 ) ( 2011 ) 847859
|
cloud service providers offer various facilities to their clients .
the clients with limited resources opt for some of these facilities
. they can outsource their bulk data to the cloud server .
the cloud server maintains these data in lieu of monetary benefits .
however , a malicious cloud server might delete some of these data to save some space and offer this extra amount of storage to another client .
therefore , the client might not retrieve her file ( or some portions of it ) as often as needed .
proofs of retrievability ( por ) provide an assurance to the client that the server is actually storing all of her data appropriately and they can be retrieved at any point of time . in a dynamic por scheme , the client can update her data after she uploads them to the cloud server .
moreover , in publicly verifiable por schemes , the client can delegate her auditing task to some third party specialized for this purpose . in this work ,
we design a publicly verifiable dynamic por scheme that is more efficient ( in terms of bandwidth required between the client and the server ) than the `` state - of - the - art '' publicly verifiable dynamic por scheme .
we also analyze security and performance of our scheme .
|
recently , antiferromagnetically coupled mixed - spin chains with an alternating array of two kinds of spins have attracted interest among researchers @xcite .
integrable models of mixed - spin antiferromagnetic chains were constructed by de vega and woynarovich @xcite and the simplest case of such chains with spins @xmath3 and @xmath4 were subsequently studied @xcite .
since these integrable models are exactly solvable , they are very useful for studying ( quantum ) statistical mechanical properties . although ferrimagnetic spin chains exhibit both ferromagnetic and antiferromagnetic features , they show some peculiar , and sometimes surprising , features uncharacteristic of either the ferromagnet or the antiferromagnet
an example being the existence of gapless excitations with very small correlation length .
it is important to understand these features more clearly . in this paper
we first present a numerical study on the quantum ferrimagnetic spin chain in magnetic field by an exact diagonalization of a finite size system .
we show that the low- and the high - field phases correspond respectively to the classical nel and the fully polarized ferromagnetic states where in the intermediate magnetic field ( @xmath1 ) , it is an xxz+h model with easy plane anisotropy @xcite , and the critical exponents are derived numerically .
then the physical properties of the quantum ferrimagnets in the presence of the uniform magnetic field , as described by a phenomenological field theory based on a continuum non - linear @xmath0 model ( nl@xmath0 m ) formulation are examined further and will explore its novel features .
when @xmath5 , the quantum ferrimagnets can be described by an easy - plane anisotropy state where the broken @xmath6 symmetry of spins corresponds to a topological spiral ( superfluid ) state .
a metastable state with a non - zero topological spin - supercurrent @xmath7 , is the superfluid state of the order parameter in xy - plane
. such possibility , may be led to the remarkable spin - dependent transport phenomenon in the ferrimagnetic chains , i.e. , the spin - supercurrent is carried collectively ( rather than by quasiparticles ) . because the spin current is non - zero when the system is in equilibrium , it flows without dissipation .
the spiral state can be characterized by a wave - vector @xmath8 , i.e. , the soliton ( saw tooth ) lattice spacing @xmath9 is a length scale through which the in - plane phase @xmath10 of the order parameter changes by @xmath11 .
hence , the spin - supercurrent is characterized by these wave - vectors , e.g. , @xmath12 . in analogy with the dissipation mechanism of the supercurrent in a one - dimensional superconductors
@xcite , at large @xmath7 , hard - axis fluctuations of the order parameter becomes unstable and a first order phase transition to a uniform in - plane state ( @xmath13 ) takes place , and the long range order of the soliton lattice destroys . by overcoming the nucleation energy barrier of the order parameter vortices , a first order phase transition from easy plane spiral state to a saturated ferromagnetic state occurs at @xmath2 , depends on @xmath7 , and @xmath8 .
this transition turns out to a continuous cross - over if the initial state of the system at @xmath14 holds the zero spin - supercurrent ( a uniform easy plane with @xmath15 ) .
the possibility of the existance of the superfluid phase and crossing onto a saturated ferromagnetic state ( the dependence of @xmath16 on @xmath8 , and @xmath7 ) are discussed in some detail .
the zero - temperature quantum ferrimagnetic chain is defined by @xmath17 where @xmath18 .
the low energy physics of the quantum ferrimagnets in the presence of the external magnetic field can be obtained by a 1-dimensional nl@xmath0 m . the external magnetic field couples with @xmath19 , the nel order parameter @xmath20 where @xmath21 ^ 2\right ) .
\label{nlsm_h}\ ] ] the first term in eq .
( [ lag_h ] ) is the usual ( dynamical ) berry s phase of a quantum ferromagnet , with @xmath22 the magnetization per unit cell ( pair of sites ) . here
@xmath23 is the topological term , @xmath24 , and @xmath25 @xcite .
it is this term that results in the ferromagnetic branch of the spin waves and corresponds to the trajectory of spin over a closed orbit on the unit sphere in the presence of a unit magnetic monopole at the center .
the contribution of the first term in @xmath26 is equivalent to the area enclosed by this trajectory and since either of the smaller or the larger enclosed areas on the unit sphere must lead to the same berry s phase , the magnetic moment per unit cell , i.e. @xmath27 , must be quantized with integral values @xcite . setting @xmath28 in action ( [ lag_h ] )
follows to the usual @xmath29 nl@xmath0 m in the magnetic field @xcite , applicable to the heisenberg antiferromagnets .
similar to zero magnetic field @xcite we can find the spin - wave modes . at @xmath30 the ferrimagnetic spin - waves consist of both gapless ( ferromagnetic ) and gapped ( antiferromagnetic ) modes , and at @xmath31 the low ( high ) energy physics of quantum ferrimagnets is effectively like that of a ferromagnet ( antiferromagnet ) which is formed by the chain of ( dimerized ) unit cells with magnetic moment @xmath32 . applying an external magnetic field , @xmath33 ,
develops a gapped ferromagnetic spin - waves ; in which case the energy cost for the ferromagnetic transitions is proportional to the zeeman splitting factor . unlike to ferromagnetic mode
, the effect of the external magnetic field is to suppress the antiferromagnetic gap
. clearly , this reveals similarities between the ferrimagnets and the integer spin heisenberg antiferromagnets in magnetic field @xcite .
the ground state of the ferrimagnet corresponds to the _ staggered _ configuration of spins , unless @xmath34 . at this point
the staggered state becomes unstable against the non - collinear _ spin - flop _ phase ( the partially polarized state ) of the spins when the spectrum becomes soft at @xmath35 .
when the external magnetic field exceeds @xmath16 , the system will be in a saturated ferromagnetic phase , with a quantized magnetization per unit cell .
the @xmath36 is obtained by using the dispersion relation of the spin waves based on the fully polarized state of the ferrimagnets .
it is the lower - bound of the external magnetic field , in the sense that the spin waves of the fully polarized state become soft at @xmath35 .
the outline of this paper is as follows : in next section we will present the numerical computations of lanczos method to compare the behavior of the correlation functions on the plateau and between two plateaux .
we also derive the critical exponents , and the effective hamiltonian for the latter region by using a quantum renormalization group ( qrg ) approach .
the main part of this paper , the spiral ( superfluid ) state of the quantum ferrimagnets is presented in section iii .
the details of the collective modes in the intermediate magnetic phase is given as the appendix in sec .
in order to have more accurate physical picture , we present the numerical results , by applying a lanczos method to each @xmath37-sector of hilbert space from @xmath38 to @xmath39 . here
@xmath40 is the number of sites that is used in the exact diagonalization .
a curve for magnetization vs. magnetic field for a chain of @xmath41-ferrimagnet has been given in ref .
an extrapolation to @xmath42 on exact diagonalization calculation shows there are two plateaux of magnetization below @xmath43 and above @xmath44 , associated with the magnetization @xmath45 and @xmath46 . here
@xmath47 is the energy gap between the ferro- and antiferromagnetic spin wave modes of this model . in fig .
[ fig2 ] the numerical results of the in - plane spin - spin correlations is presented for a point on the plateau ( at @xmath48 ) and some intermediate points between two plateaux , i.e. , @xmath49 and @xmath50 .
this exhibits the in - plane spin - spin correlation functions within the intermediate magnetic field region which falls off as power law , and manifests the critical behavior .
thus the in - plane correlation is expected to have the asymptotic form @xmath51 .
for instance we have calculated this exponent for m=@xmath52 by using data of exact diagonalization of a chain with length @xmath53 .
since there are correlations between different types of sublattices we find an exponent for each case . for correlations on sublattice a ( fig.[fig2]-(a ) ) and m=@xmath52 we obtained @xmath54 and for sublattice b ( fig.[fig2]-(b ) ) we have @xmath55 .
a similar behavior is seen for the correlations of different sublattices a and b ( fig.[fig2]-(c ) ) where @xmath56 . to have more qualitative picture of the transient region between two plateaux
, we can use a quantum renormalization group . to perform this ,
we choose two adjacent spins @xmath57 as the building block .
the block hamiltonian is then @xmath58 . for low field limit ( e.g. @xmath59 )
the lowest lying states of @xmath60 are a spin-1/2 doublet .
this comes out with an effective hamiltonian of ferromagnetic heisenberg chain with @xmath61 in magnetic field , corresponds to the @xmath48 plateau . for high field limit @xmath62 where two states @xmath63 , and @xmath64 are nearly degenerate ,
we arrive with a spin-@xmath4 antiferromagnetic @xmath65 hamiltonian .
the effective hamiltonian is @xmath66 = 7.5 cm = 9.0 cm = 7.5 cm = 4.5 cm where @xmath67 is spin-1/2 operator , @xmath68 .
it is known such model hamiltonian in the presence of a magnetic field ( @xmath33 ) has a critical line which separates the partially polarized phase from the fully polarized one @xcite .
the magnetization of the ferrimagnetic chain ( @xmath69 ) is related to the magnetization of @xmath65 model by @xmath70 which leads to the same behavior as the numerical results but with some discrepancies for the critical fields ( @xmath71 ) .
however one may continue the quantum rg procedure for the @xmath65 model and obtain that the rg flow for the partially polarized phase ( between the two plateaux ) goes to the isotropic @xmath72 fixed point @xcite .
by generalizing the nl@xmath0m@xmath73h we propose a phenomenological field theory which can present the long wave length limit of the quantum ferrimagnets within @xmath5 @xmath74 , \label{eq9}\ ] ] where we applied the spin coherent state formalism in eq.([eq9 ] ) . here
@xmath75 is the effective magnetic field . @xmath76 and @xmath77 are in - plane and out of plane spin stiffness .
@xmath78 gives an easy plane anisotropy . to study a superfluid phase , it is convenient to introduce the following variational solutions @xmath79 ) , where @xmath80 and @xmath81 . here @xmath8 is a spiral state wave vector responsible for the superfluid phase slip , and @xmath9 is the soliton lattice spacing , i.e. , the phase @xmath10 changes by @xmath11 along the chain .
clearly @xmath82 is the uniform solution ( @xmath83 is a constant ) and @xmath84 is the corresponding energy per unit length .
note @xmath85 is the momentum density conjugate to field variable @xmath10 , and the hamiltonian ( [ eq9 ] ) gives a linearly dispersing collective mode , associated with the @xmath6 symmetry breaking phase , i.e. , the superfluid phase .
it is easy to check how the classical solutions and the fluctuating modes can be derived by eq.([eq9 ] ) .
for example @xmath86 leads to @xmath87 where @xmath88 is the energy gap of the out - of - plane modes at zero wave vector .
then @xmath89 is the gauge invariance topological spin - supercurrent density carried along the @xmath90-direction .
furthermore , it is necessary to study the stability of the superfluid phase against the quantum fluctuations .
this governs with the in - plane fluctuations @xmath91 , and the out of plane fluctuations @xmath92 as one can see @xmath93 at @xmath94 .
the gapless linear superfluid mode can be obtained by @xmath95 .
the details of the collective modes calculation is presented in appendix . through out this formulation , the zero temperature phase diagram of this system can be obtained . at @xmath94 where @xmath96 the out of plane instability occurs at @xmath97 . at this point the energy gap of the @xmath98-modes vanishes and the @xmath29 symmetry of the hamiltonian is restored ( the dispersion relation of the collective modes become imaginary at small @xmath99 ) .
it follows the transition to a uniform solution must occurs at this point since @xmath100 where the linear solution can be considered as a map from the compactified physical space to the equator of the order parameter space @xmath101 .
such instability in order parameter space is induced by the local fluctuations of the order parameter , out of the equator toward the north pole of the order parameter space @xmath101 .
when @xmath7 is close to its critical values ( @xmath102 ) , the local fluctuations are very strong , and it is likely the order parameter pass the north pole , and the phase @xmath10 becomes singular , i.e. , a vortex in @xmath10 is nucleated , and the phase winding is lost .
the effect of @xmath103 is pushing the spins to be aligned along z - axis , e.g. , @xmath104 at large enough @xmath103 .
we notice if @xmath105 , system undergoes onto the saturated ferromagnetic phase by increasing the spin - supercurrent density .
this happens at @xmath106 , corresponding to @xmath107 , before the outset of the easy - plane uniform state . to study the nucleating mechanism of the vortices , we implement a mechanical analogy to the classical field theory .
this formalism has been developed by langer , and ambegaokar @xcite for one dimensional superconductors . to avoid the complexity
let us neglect @xmath77 in energy functional ( [ eq9 ] ) from now on . to start this calculation , we make the following transformations @xmath108 , @xmath109 , and @xmath110 ( where @xmath111 ) , and @xmath112 & = & \int dx \left(\frac{\rho_s^\perp}{2}\left[(\partial_x f)^2 + \frac{l^2}{f^2}\right ] \right .
\nonumber \\ & & \left .
+ \beta(1-f^2 ) - h^ * \sqrt{1-f^2 } \right ) , \label{eq10}\end{aligned}\ ] ] where @xmath113 is the momentum conjugate to @xmath10 and it is proportional to the spin - supercurrent density @xmath7 , because @xmath114 is the solution of the @xmath115/\delta\varphi=0 $ ] .
it is straightforward to show how @xmath116/\delta f=0 $ ] yields [ eq14 ] @xmath117
@xmath118 @xmath119 uniform solutions ( @xmath120 ) are one of the solutions of eqs .
( [ eq14 ] ) .
it leads to @xmath81 and @xmath121 .
the energy potential associated with the uniform solutions are depicted in fig .
[ fig_ueff ] where @xmath122 vs. @xmath123 is plotted for @xmath124 for different @xmath8 s , or equivalently for different spin - supercurrent @xmath7 .
here @xmath125 , and @xmath126 .
the effective potential , @xmath127 , has one minimum at @xmath128 ^ 2 \neq 0 $ ] if @xmath129 .
6.0 cm .5 cm this minimum energy solution is disappeared if @xmath130 , and is replaced by another minimum at @xmath131 . at @xmath132 a first order phase transition from @xmath133 onto @xmath131 takes place by nucleating of a vortex .
one can follow the transition between these minimum energy solutions by changing the parameters , @xmath8 and @xmath103 , as illustrated in fig .
[ fig_ueff ] .
the saddle point solutions can be derived by setting @xmath134 at the extremum of @xmath135 .
it follows at @xmath136 , the local minima at finite @xmath137 is disappeared and @xmath104 becomes a unique minima of @xmath138 .
moreover , the cross - over from the uniform state [ a broken @xmath6 symmetry state with zero spin - supercurrent ] to a saturated ferromagnetic state occurs at @xmath139 ( @xmath140 ) , consistent with the exact diagonalization results ( @xmath44 ) . obviously at @xmath14 and @xmath141 ,
the sign of curvature of @xmath127 changes , and therefore @xmath142 can be considered as the saddle point of @xmath127 , at which the spiral state becomes unstable against the hard - axis fluctuations and a transition to a uniform solution occurs .
this is seen if the spin - supercurrent @xmath7 arrives to its critical value , @xmath102 . as was mentioned earlier
, the mechanism for such transition is nucleating of a vortex in the order parameter space . for a given @xmath8 and @xmath103 ,
it is straightforward to show @xmath143 .
recalling @xmath144 leads to @xmath145 ( @xmath146 ) which is the spin - supercurrent density and @xmath147 is the classical solution of eq.([eq14 ] ) .
the dependence of @xmath148)$ ] with respect to @xmath8 for various @xmath103 is illustrated in fig .
it follows that @xmath7 vanishes at @xmath149 where the system crosses to the saturated ferromagnetic phase .
the maximum spin - supercurrent @xmath150 , which can pass through the system at finite wave vector @xmath8 ( and @xmath151 ) , is a monotonically decreasing function of @xmath103 .
6.0 cm .5 cm in conclusion , we have predicted the intermediate magnetic field region of ferrimagnets can support the dissipationless flow of the spin - supercurrents , which may be observed by the advance techniques of _ spintronics_. the idea of the collective spin - supercurrent transport which has been presented in this paper is completely general , and can be applicable to any 1-dimensional spin system with easy plane anisotropy .
the intermediate magnetic field phase of the ferrimagnets is one example .
other example of this kind is the _
isospin_-supercurrent transport of the quantum hall bars in the bilayer electron systems .
here we present the detailed calculation of the collective modes in the intermediate magnetic field phase .
the effective hamiltonian in terms of the in - plane and the out of plane fluctuations is given by @xmath152 & = & \frac{1}{2 } \sum_q \varphi(-q ) k_{\varphi\varphi}(q ) \varphi(q ) \nonumber \\ & & + \frac{1}{2}\sum_q n_z(-q ) k_{zz}(q ) n_z(q),\end{aligned}\ ] ] where @xmath153 and @xmath154 .
the equation of motion for the field variables @xmath10 , and @xmath85 can be obtained by the hamilton equations ( @xmath155 and @xmath156 ) [ jj ] @xmath157 @xmath158 where @xmath85 is the momentum density associated with the field variable @xmath10 . in the spin coherent representation , @xmath85 and @xmath159 are given by @xmath160 , and @xmath161 respectively . then it is easy to check the transformation @xmath162 can be obtained by canonical quantization , i.e. , @xmath85 can be considered as the momentum density conjugate to field variable @xmath10 . making the derivative with respect to time we find the equation of motion :
@xmath163 , where @xmath164 .
this clearly gives @xmath165 eqs.([jja]-[jjb ] ) are similar to the coupled josephson junction relations in superconductivity .
we are grateful to homayoun hamidian who was involved in the early stage of this study .
useful conversation with allan macdonald , miguel martin - delgado and kieran mullen is acknowledged .
work at the university of oklahoma was supported by the nsf under grant no .
eps-9720651 and a grant from the oklahoma state regents for higher education .
a. k. kolezhuk , h .- j .
mikeska , and s. yamamoto , phys .
b * 55 * , r3336 ( 1997 ) ; s. k. pati , s. ramasesha , and d. sen , phys . rev .
b * 55 * , 8894 ( 1997 ) ; s. brehmer , h .- j .
mikeska , and s. yamamoto , j. phys . : condens .
matter * 9 * , 3921 ( 1997 ) ; s. k. pati , s. ramasesha , and d. sen , j. phys .
matter * 9 * , 8707 ( 1997 ) ; g .- s .
tian , phys .
b * 56 * , 5355 ( 1997 ) ; s. yamamoto and t. fukuii , phys . rev .
b * 57 * r14008 ( 1998 ) ; n. b. ivanov , phys . rev .
b * 57 * r14024 ( 1998 ) ; k. maisinger , u. schollwck , s. brehmer , h. j. mikeska , and shoji yamamoto , , r5908 ( 1998 ) . the magnetization curve of a 1d ferrimagnets can be obtained by setting @xmath166 in the sba configuration of two - leg ladder , presented in a. langari and m. a. martin - delgado , , 11725 ( 2000 ) , and normalizing the magnetic field by 2 .
|
we study the phase diagram of one - dimensional quantum ferrimagnets by using a numerical exact diagonalization of a finite size system along with a field - theoretical non - linear @xmath0 model of the quantum ferrimagnets at zero temperature and its effective description in the presence of the external magnetic field in terms of the quantum xy - model .
the low- and the high - field phases correspond respectively to the classical nel and the fully polarized ferromagnetic states where in the intermediate magnetic field ( @xmath1 ) , it is an xxz+h model with easy plane anisotropy , which possess the spiral ( superfluid ) states that carry the dissipationless spin - supercurrent .
we derive the critical exponents , and then will study the stability of the xy spiral state against these spin - supercurrents and the hard axis fluctuations .
we will show a first order phase transition from the easy plane spiral state to a saturated ferromagnetic state occurs at @xmath2 if the spin - supercurrent reaches to its critical value . 2
|
planet formation theories thus far extract much of their information from the known transiting exoplanets which are largely in the short - period regime .
this is because transit surveys which have provided the bulk of the transiting planet discoveries , such as superwasp @xcite and hatnet @xcite , are biased towards this region .
the two planets which contribute to the sample of intermediate to long - period transiting exoplanets are : hd 17156b @xcite and hd 80606b @xcite , the latter of which exhibits both a primary transit and secondary eclipse . in both of these cases ,
the detection was largely due to the inflated transit / eclipse probability caused by their extreme orbital eccentricities .
both of these were observed to transit through predictions based upon their radial velocity data .
planetary orbits may be considered in three basic categories : short - period ( @xmath0 days ) planets , intermediate to long - period high - eccentricity ( @xmath1 ) planets , and intermediate to long - period low - eccentricity planets .
the first type of planet are the focus of current studies , and we are beginning to gain insight into the second type ( hd 80606b for example ) .
exploration into the structure of those planets occupying the second and third orbit types will require probing parameter - space beyond that currently encompassed by the known transiting exoplanets .
the science objectives of such an exploration include understanding how planetary properties , such as average planet density , vary with periastron distance as well as providing the first observational data for exoplanet models with low incident stellar radiation .
recent observations of hd 80606b by @xcite and @xcite suggest a spin - orbit misalignment caused by a kozai mechanism ; a suggestion which could be investigated in terms of period and eccentricity dependencies if more long - period transiting planet were known .
figure [ peri ] shows the distribution of periastron distances for the known exoplanets using data from jean schneider s extra - solar planets encyclopaedia .
also shown as a shaded histogram is the same distribution for the known transiting exoplanets .
clearly the periastron distribution of the known transiting planets does not accurately represent the distribution of the entire sample of known exoplanets .
thus far , our picture of exoplanets is based on the planets with super - heated atmospheres through either short - period orbits or intermediate - period eccentric orbits .
figure 1 shows how our view of planetary properties is highly biased towards planets of short periastron distance , thus demonstrating the need for detecting transits of exoplanets at larger distances in order to study atmospheres with greatly reduced flux from the parent star .
long - period transits will give us valuable insight into the structure of exoplanets that are more similar to those in our own solar system .
additionally , host stars of long - period planets discovered using the radial velocity technique tend to be very bright , and potentially one of these could be the brightest star with a transiting jupiter - mass planet .
the brightness of such host stars can faciliate atmospheric studies via transmission spectra @xcite and thermal phase variations @xcite .
the long period typically also translates to a longer transit duration , allowing significantly more signal to be collected when trying to probe the atmospheric composition of these objects with high - resolution spectrographs . since most bright late - f , g ,
k type stars have already been searched for short - period jupiter - mass planets , the long period hosts are the obvious sample to now search since their relative brightness and longer transit times are both very helpful in attempting to probe the exoplanetary atmosphere . it is clear that the discovery of long - period transiting planets is very important .
they are , however , difficult to detect in ground - based transit surveys due to the the presence of correlated ( red ) noise @xcite and effects of the observational window function @xcite .
though combining data from various on - going transit surveys to search for transiting long - period planets will improve detectability @xcite , it is easier to discover transits among the large number of long - period exoplanets already known from radial velocity surveys . based upon monte - carlo simulations of the transit probabilities , it is expected that several of the known long - period planets should transit their parent stars due to the enhanced probabilities produced by eccentric orbits @xcite .
there have been suggestions regarding the strategy for photometric follow - up of these radial velocity planets at predicted times of primary transit @xcite and secondary eclipse @xcite , and the instruments which could be used for such surveys @xcite .
here we describe a methodology through which to search for transits of known planets . in section 2
we calculate transit windows for a large selection of the known exoplanets and show the impact of additional radial velocity measurements on refining the transit ephemerides .
section 3 describes the techniques through which optimal target selection and observations of the transit window can be achieved .
the methods described here have been successfully tested and implemented by the transit ephemeris refinement and monitoring survey ( terms ) .
the transit window as described here is defined as a specific time period during which a complete transit ( including ingress and egress ) could occur for a specified planet .
the limiting factor for successfully observing a known exoplanet host star during the predicted transit window is often the precision of the transit ephemeris .
the quality of the transit ephemeris is primarily determined by ( a ) the uncertainties associated with the fitted orbital parameters , and ( b ) the time elapsed since the most recent radial velocity data was acquired .
acquiring new high precision radial velocity data can easily mitigate both effects . with a prompt photometric observing strategy after the orbital parameters have been revised
, one can maximize the chances of being able to obtain complete coverage of the transit observing window and thus either confirm or rule out the transiting nature of the planet .
the orbital parameters measured from fitting the radial velocity data of a planet are sufficient for calculating a transit ephemeris .
the predicted time of mid - transit can be calculated by using kepler s equations .
firstly , the eccentric anomaly is calculated from the following @xmath2 where @xmath3 is the orbital eccentricity and @xmath4 is the true anomaly . as described in @xcite ,
the time of transit mid - point will occur when @xmath5 , where @xmath6 is the argument of periastron . substituting this for the true anomaly in equation [ eccanom ]
thus yields the eccentric anomaly at the point of predicted transit . the mean anomaly , @xmath7 , which defines the time since last periapsis in units of radians , is then computed by @xmath8 which can be converted to regular time units using @xmath9 where @xmath10 is the orbital period .
the predicted mid - point of primary transit can then be calculated using @xmath11 where @xmath12 is the time of periastron passage and the term of @xmath13 incorporates the number of complete orbits which have transpired since @xmath12 . [ cols="^,^ " , ]
amongst the southern hemisphere rv planets , there are few that have been adequately monitored photometrically to confirm or rule out planetary transits . by calculating transit ephemerides for these planets and designing an efficient observing program , it is possible to examine a large subset of these planets with a relatively small amount of observing time on a 1.0m - class telescope . by applying strict criteria on the predicted transit properties of the targets ,
we are able to produce a robust selection which yields the most promising targets on which one can place transit constraints .
these criteria and general observing strategy considerations are described here . for each observing run
, the experimental design constitutes the selection of targets which meets the necessary criteria ( described below ) for successful observations to be undertaken .
the primary challenge is to match transit windows with observability for each target which is not a trivial task .
here we describe the minimum criteria that must be met for each target .
the first step is to select all known radial velocity targets for which an estimate of the stellar radius , either from measurement or models , is available for calculating the transit depth and the geometric transit probability .
the probability that is most important for the target scheduling is the transit detection probability which is a combination of the geometric transit probability and the fraction of the transit window during which the target is observable . to first order ,
this is a straight multiplication but will depend upon the probability distribution of the predicted transit mid - point , as discussed in section [ coverage ] .
based upon the photometric precision of the experimental system , the predicted transit depth can be used to exclude targets whose depth is too low , particularly giant host stars .
the steps thereafter depend upon how the observing time is allocated ; a fixed time - slot ( such as noao time ) , or queue - scheduled / service time ( such as that used by @xcite ) . for a fixed time - slot ,
the essential steps are : 1 . for 0.91.0 m class telescopes , stars brighter than @xmath14 often need to be excluded unless the telescope has the options of aperture diaphragms or neutral density filters
. 2 . include only those stars whose airmass is less than @xmath15 for at least 3 hrs during the night . a transit window of 3 hours is the likely minimum transit window available and so the visibility of the target will still be useful if the transit window happens to largely coincide .
3 . for each target that passes the visibility and brightness criteria ,
the transit ephemeris is checked and transit times are noted for those predicted transits which fall on dates during the run .
4 . for each date during the run , the ut times of the transit are checked and transit windows which occur during the day are rejected .
transit windows which do not coincide with the observable hours are also rejected .
the schedule for each night is considered in terms of the transit detection probability ( described above ) and the targets are ranked in descending order of the probability . if one seeks to concentrate the investigation on long - period planets , then the planets should be ranked by the observability of the transit windows since the geometric transit probability will dominate the transit detection probability for short - period planets .
queue - scheduled observations essentially allow an opportunity to target long - period planets whose transit windows occur far less frequently than those usually monitored during pre - allocated observing runs . for queue - scheduled observations the steps are as follows : 1 .
rank the exoplanets from long - period to short - period and keep only those whose transit probability exceeds the geometric transit probability for a circular orbit ( see @xcite ) .
2 . examine the the transit window for each planet and reject those for which the transit windows are excessively long ( for example , greater than @xmath16 days ) .
3 . investigate the visibility of each target from the observing site and reject the targets for which there is a mis - match between the occurance of the transit window and the time the target is observable .
ideal targets are those for which the transit window occurs during a single night and the target is up all night . for targets which have transit windows spanning multiple nights , the decision will be based upon the value of the target in terms of the transit detection probability .
these steps are almost the reverse of the steps recommended for fixed time - slot observing runs . since queue - scheduled observations allow for targeted observations of rare transit windows for long - period planets , they are thus those which will yield the highest success for these high - risk / high - return targets .
the difficulties in establishing an optimal observing schedule not withstanding , there are further considerations that one needs to take into account when planning observations .
one of these is the decision about which parts of the transit window to monitor if the window spans more than one night ( and often several nights for long - period planets ) , especially if one or more of those nights competes with other favourable targets .
so far we have assumed that all parts of the transit window are equally significant in the likelihood of a transit being observed or ruled out .
this is generally not true but depends upon the probability distribution of the orbital parameter uncertainties .
for example , we can suggest an empirical model whereby we assume gaussian uncertainties for the fit parameters , which in turn assumes gaussian noise in the radial velocity measurements . ) , assuming that the uncertainties in the orbital parameters are gaussian.,width=309 ] figure [ winprob ] shows a gaussian probability distribution for the predicted location of the transit mid - point for a planet which has a transit window size 5% of the orbital period . if one is able to monitor the target for only half of the transit window , then choosing this range to be centered on the predicted transit mid - point ( shown as the shaded region in figure [ winprob ] )
will account for 38% of the area under the probability distribution , as opposed to 30% for the remainder of the time within the transit window . in reality
, the uncertainties associated with the orbital parameters will have a more complex distribution due to systematic noise components .
their distributions may be close to that described by gaussian or poisson statistics , but can be determined empirically through monte - carlo simulations which randomize the sequence of the residuals on the radial velocity measurements and redetermining the orbital fit ( see @xcite and @xcite for more details ) .
in addition , the probability distribution for a particular parameter is usually non - symmetric in nature , as seen in a @xmath17 map of the parameter whilst keeping the other parameters fixed ( for example , see @xmath17 maps by @xcite ) , but can be approximated as symmetric at the @xmath18 level .
whatever the distribution , the measurements per unit time will be more valuable the closer they are to the predicted mid - point .
the caveat to this is when the predicted transit duration is larger than the time for which the target is observable during the night , since it is important to observe either ingress or egress for relative photometry .
thus , if the transit duration is greater than the observing window then the optimal approach is to observe at @xmath19 .
the coverage of the transit window can be increased through the use of telescopes adequately separated in longitude .
such networks are already in existence ( e.g. , las cumbres observatory global telescope ( lcogt ) network ) and collaborations for the follow - up of gamma - ray bursts and microlensing events are quite common .
queue scheduling of observations is particularly useful for the rare transit windows of long - period planets .
this kind of observing is available , for instance , to member consortiums who utilise the service time of the observing with small and moderate aperture research telescope system ( smarts ) . for the especially bright targets which will not only saturate typical 1.0m - class telescopes but
whose field - of - view will also be devoid of comparison stars , a solution is to use the microvariability and oscillations of stars ( most ) satellite @xcite .
the most satellite has demonstrated photometric precision of a few parts per million which is sufficient for detecting transit signatures due to planets orbiting bright giant stars .
since most is space - based , this would also allow complete coverage of the transit window without the need for coordinated ground - based observations using different telescopes ( an additional source of red noise ) .
provided the transit window can be provided with sufficient accuracy , this would be an excellent use of the most satellite s capabilities .
many of the known radial velocity planets have yet to be surveyed for transit signatures .
the detection of a transit for the intermediate to long - period planets would add enormously to our knowledge of planetary structure and , in particular , how the structure varies with semi - major axis and periastron distance .
the advantages of targeting long - period radial velocity planets are the brightness of the host stars and the prior knowledge of the planetary orbital parameters . however , a major challenge of monitoring the host stars at predicted transit times is that many transit windows have deteriorated over time , such that the telescope time required renders attempts to do so impractical .
we have shown through calculations for 245 of the known exoplanets how the size of the transit window varies with period and geometric transit probability .
the large uncertainties associated with the transit mid - point for the long - period planets is dominated by the uncertainties in the period and time of periastron passage estimated from the discovery data
. we demonstrated , using the examples of hd 190228 and hd 231701 , that a handful of carefully timed additional measurements can vastly improve the size of the transit window and thus bring the monitoring of the window into the reach of ground - based programs .
the difficulties involved in the observing schedule largely result from matching transit windows with the observability of the targets , particularly for long - period planets whose transit windows are widely spaced .
we have described a planning strategy which will make optimal use of both pre - allocated and service telescope time , also noting the advantages of both longitude coverage and space - based observations .
it is important to consider for the scheduling that the central part of the transit window can be significantly more valuable than the wings of the window , depending upon the nature of the orbital parameter uncertainties .
the described techniques and science goals are currently being undertaken and investigated by the transit ephemeris refinement and monitoring survey ( terms ) .
note that the observations from this survey will lead to improved exoplanet orbital parameters and ephemerides even without an eventual transit detection for a particular planet .
the results from this survey will provide a complimentary dataset to the fainter magnitude range of the kepler mission @xcite , which is expected to discover many transiting planets including those of intermediate to long - period planets .
the authors would like to thank steven berukoff for several useful suggestions .
this research has made use of the nasa / ipac / nexsci star and exoplanet database , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
bakos , g. . ,
lzr , j. , papp , i. , sri , p. , green , e.m . , 2002 , pasp , 114 , 974 barbieri , m. , et al .
, 2007 , a&a , 476 , l13 barnes , j.w . , 2007 , pasp , 119 , 986 borucki , w.j . , et al . , 2009 , science , 325 , 709 burrows , a. , sudarsky , d. , hubeny , i. , 2006 , apj , 650 , 1140 cumming , a. , 2004 , mnras , 354 , 1165 fischer , d.a . , et al . , 2007 , apj , 669 , 1336 fleming , s.w . , kane , s.r . , mccullough , p.r . , chromey , f.r . , 2008 , mnras , 386 , 1503 ford , e.b . , 2005 , apj , 129 , 1706 ford , e.b . , 2008 , apj , 135 , 1008 gillon , m. , 2009 , mnras , submitted ( arxiv:0906.4904 ) kane , s.r . , 2007 , mnras , 380 , 1488 kane , s.r . ,
schneider , d.p .
, ge , j. , 2007 , mnras , 377 , 1610 kane , s.r . , von braun , k. , 2008 , apj , 689 , 492 kane , s.r . , ford , e.b .
, ge , j. , 2008 , iaus , 249 , 115 kane , s.r . , von braun , k. , 2009 , pasp , in press ( arxiv:0908.0519 ) kane , s.r . , mahadevan , s. , cochran , w.d . , street , r.a . ,
sivarani , t. , henry , g.w . ,
williamson , m.h .
, 2009 , apj , 692 , 290 knutson , h.a . , et al . , 2007 , nature , 447 , 183 laughlin , g. , deming , d. , langton , j. , kasen , d. , vogt , s. , butler , p. , rivera , e. , meschiari , s. , 2009 , nature , 457 , 562 lpez - morales , m. , 2006 , pasp , 118 , 716 matthews , j.m .
, kusching , r. , guenther , d.b . ,
walker , g.a.h . ,
moffat , a.f.j . ,
rucinski , s.m . ,
sasselov , d. , weiss , w.w . , 2004 , nature , 430 , 51 moutou et al . , 2009 ,
a&a , 498 , l5 perrier , c. , sivan , j .-
p . , naef , d. , beuzit , j.l . ,
mayor , m. , queloz , d. , udry , s. , 2003 , a&a , 410 , 1039 pollacco , d.l .
, et al . , 2006 , pasp , 118 , 1407 pont , f. , zucker , s. , queloz , d. , 2006 , mnras , 373 , 231 pont , f. , et al . , 2009 , a&a , 502 , 695 redfield , s. , endl , m. , cochran , w.d . ,
koesterke , l. , 2008 , apj , 673 , l87 snellen , i.a.g . ,
albrecht , s. , de mooij , e.j.w .
, le poole , r.s . , 2008 , a&a , 487 , 357 von braun , k. , kane , s.r . , ciardi , d.r . , 2009 ,
apj , in press ( arxiv:0907.1614 ) walker , g.a.h .
, et al . , 2003 , pasp , 115 , 1023 wright , j.t . , howard , a.w . , 2009 , apjs , 182 , 205
|
transiting planet discoveries have yielded a plethora of information regarding the internal structure and atmospheres of extra - solar planets .
these discoveries have been restricted to the low - periastron distance regime due to the bias inherent in the geometric transit probability .
monitoring known radial velocity planets at predicted transit times is a proven method of detecting transits , and presents an avenue through which to explore the mass - radius relationship of exoplanets in new regions of period / periastron space . here
we describe transit window calculations for known radial velocity planets , techniques for refining their transit ephemerides , target selection criteria , and observational methods for obtaining maximum coverage of transit windows .
these methods are currently being implemented by the transit ephemeris refinement and monitoring survey ( terms ) .
|
abundance determinations of stars in galactic globular clusters can provide valuable information about important astrophysical processes such as stellar evolution , stellar structure , galactic chemical evolution and the formation of the milky way .
surface stellar abundances of c , n , o , and often na , mg , and al are found to be variable among red giants within a globular cluster .
the physical process responsible for these star - to - star element variations is still uncertain ( see the reviews of kraft 1994 and pinsonneault 1997 , as well as cohen _
et al . _
2001 , paper i ) . in order to study the origin of the star - to - star abundance variations
, we have started a program to determine chemical abundances of the nearer galactic globular cluster stars .
@xcite presents the sample of stars in m71 , the nearest globular cluster reachable from the northern hemisphere , and the atmosphere parameters of the program stars .
our sample includes stars over a large range in luminosity : 19 giant stars , 3 horizontal branch ( hb ) stars , and 3 stars near the main sequence turnoff , in order to study in a consistent manner red giants , horizontal branch stars , and stars at the main sequence turnoff . our second paper ( * ? ?
* paper ii ) discusses the iron abundance in m71 .
we found that the [ fe / h ] abundances from both fe i ( [ fe / h ] = @xmath4 ) and fe ii ( [ fe / h ] = @xmath5 ) lines agree with each other and with earlier determinations @xcite .
we also found that the [ fe / h ] obtained from fe i and fe ii lines is constant within the rather small uncertainties for this group of stars over the full range in effective temperature ( @xmath2 ) and luminosity . in this third paper of this series
, we present our results for abundances of 23 atomic species in our sample of m71 stars .
the abundance analysis is done using a current version of the lte spectral synthesis program moog @xcite .
a line list specifying the wavelengths , excitation potentials , @xmath6 values , damping constants , and equivalent widths for the observed lines is required .
the provenance of the equivalent widths , @xmath6 values and damping constants is discussed below .
in addition , a model atmosphere for the @xmath2and surface gravity ( log(@xmath7 ) ) appropriate for each star and a value for the microturbulent velocity ( @xmath8 ) are also required for the abundance analysis .
we use the grid of model atmospheres from @xcite with a metallicity of [ fe / h ] = @xmath90.5 dex , based on earlier high dispersion iron abundance analysis of m71 ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* paper ii ) .
@xmath2 and log(@xmath7 ) are derived from the broad - band photometry of the stars as described in paper i. the photometric @xmath2 has an error of @xmath1075 k for giants and @xmath10150 k for dwarfs and log(@xmath7 ) has an error of @xmath100.2 dex .
the microturbulent velocity is derived as described in paper ii ; @xmath8 has an error of @xmath100.2 km s@xmath11 .
table [ tab1 ] , reproduced from paper ii , lists the stellar parameters of our sample of m71 stars . the search for absorption features present in our hires data and the measurement of their equivalent width ( @xmath12 )
was done automatically with a fortran code , ewdet , developed for this project .
details of this code and its features are described in paper ii . the line list identified and measured by ewdet is then correlated to the list of suitable unblended lines with atomic parameters to specifically identify the different atomic lines .
the list of unblended atomic lines was created by inspection of the spectra of m71 stars , as well as the online solar spectrum taken with the fts at the national solar observatory of wallace , hinkle & livingston ( 1998 ) and the set of solar line identifications of moore , minnaert & houtgast ( 1966 ) . in paper ii , we derived the @xmath13@xmath12 relation of the fe i lines of `` the weak line set '' ( fe i lines within two sigma levels of the @xmath13@xmath12 fit , @xmath12 @xmath14 60 m , and errors less than a third of the @xmath12 ) .
we used these @xmath13@xmath12 relations to determine `` the good line set '' ( lines with errors less than a third of the @xmath12 and with @xmath12computed from the derived @xmath13@xmath12 relations ) . the @xmath12 of the lines presented in this paper
are also determined using the fit to the @xmath13@xmath12 relation of the fe i lines of `` the good line set '' , except for the c i , o i and ca i lines , and for the elements that show hyperfine structure splitting ( sc ii , v i , mn i , co i , cu i , and ba ii ) . the equivalent widths of the c i and o i lines were measured by hand , since thermal motions become important at the low atomic weights of these elements and the @xmath13@xmath12 relations derived for fe i lines may no longer be valid . for ca i lines and the lines of elements that show hyperfine structure splitting , we used the equivalent widths measured automatically by ewdet , but did not force them to fit the fe i @xmath13@xmath12 relationship due to the probable different broadening mechanisms .
many of the ca i lines were strong enough to be on the damping part of the curve of growth .
the @xmath12 used in the abundance analysis are listed in table [ tab2 ] ( available electronically ) , which also includes the @xmath12 for the fe i and fe ii lines used in paper ii .
transition probabilities for the lines used in this analysis were obtained from the nist atomic spectra database ( nist standard reference database # 78 , see @xcite ) when possible .
nearly 80% of the lines selected as suitable from the hires spectra have transition probabilities from the nist database . for the remaining lines the @xmath6 values come from the inverted solar analysis of @xcite , corrected by the factors listed in table [ tab3 ] which are needed to place both sets of transition probabilities onto the same scale .
the correction factor was computed as the mean difference in log(@xmath6 ) between the nist and solar values for the lines in common , which number is given in column 2 of table [ tab3 ] .
elements not listed in table [ tab3 ] have transition probabilities from the nist database for all their lines utilized here .
six elements show hyperfine structure splitting ( sc ii , v i , mn i , co i , cu i , and ba ii ) . the corresponding hyperfine structure constants were taken from @xcite .
most of the na i and ca i lines are strong enough for damping effects to be important . for na i the interaction constants , @xmath15 , of the van de waals broadening were taken from the solar analysis of @xcite .
@xcite studied collisional broadening of 17 ca i lines . comparing their experimental results and the predicted values of @xmath15
obtained using the unsld approximation , we found that the experimental @xmath15 are about 10 times larger than the unsld @xmath15 .
thus for the ca i we used the experimental @xmath15 from @xcite when available , otherwise we use 10 times the unsld approximation
. the empirical values of @xmath15 for al i and mg i from @xcite and @xcite , respectively , are also used .
we used 4 times times the unsld approximation for those al i lines without empirical damping constants .
for the lines of all other ions we set @xmath15 to be twice the unsld approximation as was done in paper ii for the fe i lines following @xcite .
we need to establish the solar abundances corresponding to our adopted set of @xmath6 values and damping constants .
solar abundance ratios were computed using our compilation of atomic parameters , the kurucz model atmosphere for the sun @xcite and the list of equivalent widths from @xcite .
the results are listed in table [ tab4 ] . the o abundance from the permitted lines
are corrected by a factor of 0.35 dex ( see below ) .
there is a general agreement with the meteoric solar abundance ratios from @xcite .
the difference between our solar abundances and the meteoric solar abundances from @xcite is listed in column 5 of table [ tab4 ] .
this difference is within the standard deviation of our own measurements , with the exception of [ ca / fe ] , which is our most deviant abundance ratio .
the difference we found is almost the same as the correction factor applied to the solar @xmath6 values , listed in table [ tab3 ] .
we use these solar abundance ratios , derived from our choice of atomic line parameters , to compute the abundance ratios for our sample of m71 stars .
the non - lte treatment of the oxygen permitted lines is discussed in [ abun ] .
the k i resonance lines are strongly affected by non - lte effects in the sun @xcite .
@xcite carried out statistical equilibrium calculations for the k i line 7699 , the only line used in our analysis , for metal poor stars .
we applied a non - lte correction to our results of k abundance following @xcite .
the smallest correction applied was @xmath90.12 dex for the cool giant stars , and the largest correction was @xmath90.7 dex for the hb stars . without these corrections , a very strong dependence of k abundance on @xmath2 ( equivalent to luminosity ) was seen .
the aluminium lines are also affected by non - lte in metal poor stars @xcite .
unfortunately , the statistical equilibrium calculations of @xcite included only dwarf stars .
their non - lte corrections increase with decreasing metallicity , and are larger for the 3961 line than for the 6697 doublet used in our analysis .
al and all other elements were treated assuming lte .
given the stellar parameters from paper i , we determined the abundances using the equivalent widths obtained as described above . the abundance analysis is done using a current version of the lte spectral synthesis program moog @xcite .
we employ the grid of stellar atmospheres from @xcite to compute the abundances of c , o , na , mg , al , si , k , ca , sc , ti , v , cr , mn , co , ni , cu , zn , y , zr , ba , la , and eu using the four stellar atmosphere models with the closest @xmath2 and log(@xmath7 ) to each star s parameters . the abundances were interpolated using results from the closest stellar model atmospheres to the appropriate @xmath2 and log(@xmath7 ) for each star .
we also determine the abundance of li using a synthesis of the spectra in the area of the doublet at 6707 , which , since it is not well resolved , is considered as only one line in table [ tab5a ] .
the li abundance is given as log @xmath16(li)=log(@xmath17)+12.0 , where @xmath18 is number of atoms .
in this case , we used the stellar model atmosphere from @xcite with the closest @xmath2 and log(@xmath7 ) to each star s parameters .
the abundance ratios , with the exception of [ c / fe ] , [ o / fe ] , [ si / fe ] and [ zn / fe ] are computed using the iron abundance from fe i lines of paper ii as slightly updated in table [ tab1 ] , and our solar abundance ratios from table [ tab4 ] . given their high excitation potentials , the abundance ratios for the c i , si i , and zn i lines were computed using the [ fe / h ] from fe ii lines . in the @xmath2 range of our sample of stars ,
most of the iron is in the form of fe ii and most of the oxygen is in the form of o i , so both species behave similarly for small changes in the atmospheric parameters .
for this reason , we computed the abundance ratio of o using the fe ii lines as well .
the computed abundance ratios are listed in tables [ tab5a ] - [ tab5e ] .
there are 11 stars with equivalent widths measured for both the two forbidden and the permitted o lines .
the difference of the oxygen abundance ratio from forbidden and permitted lines for those 11 stars and the sun is plotted against @xmath2 in figure [ ox_teff ] . a clear trend with @xmath2 is observed , which may come from the different excitation potential of the forbidden and the permitted lines or from non - lte effects on the permitted lines .
we tried applying the non - lte corrections suggested by @xcite and by @xcite to the permitted lines , but the observed @xmath2 trend becomes even steeper .
@xcite and @xcite , among others , discuss the validity of the different oxygen abundance indicators . since the forbidden lines are usually considered to give more reliable abundances ( but see
* ; * ? ? ?
* ) , we corrected the abundance ratio from the permitted lines by the amount given by the least squares fit shown in figure [ ox_teff ] .
the final [ o / fe ] listed in table [ tab5a ] is the average for each star of the results from the forbidden and the corrected permitted lines .
note that a correction of 0.35 dex , which corresponds to the correction at the temperature of the sun , was applied to the abundances deduced from the permitted lines of o i in the sun to compute its [ o / fe ] in the same way as for our m71 sample of stars .
the abundance ratios ( absolute abundance for li ) for each star in our m71 sample are plotted against the photometric @xmath2 in figures [ li1 ] to [ neutron ] .
the error bars shown in figures [ li1 ] to [ neutron ] correspond to the standard deviation of results of different atomic lines divided by the square root of the number of lines used for each star .
the solid line , shown in figures [ light ] to [ neutron ] , is a linear fit weighted by the errors of the respective abundance ratio versus @xmath2 .
the dashed line shown in these figures indicates the mean abundance ratio and its respective error plotted as an error bar at 3925 k. the error in the mean abundance ratio corresponds to the standard deviation within our sample of stars divided by the square root of the number of stars for which an abundance was derived for that ion .
we estimate the sensitivity of the abundances with respect to small changes in the equivalent widths ( synthesis for li ) and the stellar parameters in four cases 4000/1.0/1.4 , 4250/1.0/1.4 , 5000/2.5/1.0 and 5500/4.0/0.6 , where the three numbers correspond to @xmath2/log(@xmath7)/@xmath8 .
the case 4000/1.0/1.4 has been computed only for elements with high excitation lines , which are more sensitive at lower temperatures .
the stellar parameters of these cases span the relevant range of atmospheric parameters for our m71 sample .
we estimated the error in the @xmath12 to be 10% for all the lines .
the error of the synthesis of the li doublet is estimated to be @xmath100.1 dex .
the results are listed in table [ tab6 ] , where the range adopted for each parameter is representative of its uncertainty . because of the high excitation of the c i lines studied here , this is the ion included in our analysis whose derived abundance is most sensitive to @xmath2 .
[ ca / h ] also has a sensitive dependence on @xmath2 and on @xmath8 , because the ca i lines are all rather strong and have large damping constants .
the mean abundance ratios and their errors are listed in table [ tab7 ] .
the statistical error , @xmath19 , corresponds to the standard deviation of sample of stars divided by the square root of the number of stars and it is a measure of the scatter of the abundance ratio in the sample of m71 stars . in order to quantify the abundance ratio variations within our sample of m71 stars we have to compare the measure of the scatter with the predicted error from the stellar parameters and the measurement of the @xmath12 ( or synthesis for li ) .
we estimated the predicted error , @xmath20 , using the following equation : @xmath21 ) = & \delta ( x : w_{\lambda})^{2}/n_{lines}(x)+\delta ( fe : w_{\lambda})^{2}/n_{lines}(fe ) + \\ & [ \delta ( x : t_{eff } ) - \delta ( fe : t_{eff})]^{2 } + \\ & [ \delta ( x:{\rm log}(g ) ) - \delta ( fe:{\rm log}(g))]^{2 } + \\ & [ \delta ( x:\xi ) - \delta ( fe:\xi)]^{2 } + \\ & [ \delta ( x:{\rm [ fe / h ] } ) - \delta ( fe:{\rm [ fe / h]})]^{2 } \end{aligned}\ ] ] where @xmath22@xmath12 ) , @xmath22@xmath2 ) , @xmath22log(@xmath7 ) ) , @xmath22@xmath8 ) , and @xmath22[fe / h ] ) are listed in columns 2 , 3 , 4 , 5 and 6 of table [ tab6 ] .
@xmath23 is the number of lines used to compute the abundances , @xmath24 denotes the element under consideration , and @xmath25 denotes either fe i or fe ii , whichever was used to compute the abundance ratio .
our @xmath20 ignores covariance among the error terms , which is discussed in detail by @xcite .
she shows that these additional terms are fairly small , and will be even smaller in our case , as we have determined log(@xmath7 ) using isochrones rather than through ionization equilibria ( see paper i ) .
the general small trends seen in figures [ li1 ] to [ neutron ] of [ x / fe ] slightly increasing toward cooler @xmath2may result from ignoring the covariance terms ( see johnson 2002 ) .
the predicted errors for each ion are listed in column 4 of table [ tab7 ] .
the maximum abundance trend over the relevant @xmath2 range for each element , @xmath26 , is also listed in column 6 of table [ tab7 ] .
this parameter , which is not sensitive to star - to - star scatter abundance variations for stars at a given evolutionary state , is the slope of the linear fit of the abundance ratio vs. @xmath2 times the range in @xmath2 ; its error is the error in the slope time the range in @xmath2 covered by the sample of stars in which the ion of interest was observed .
the values of @xmath26 for essentially all elements observed are gratifyingly small , providing evidence to support many of the assumptions made in the course of this analysis , such as that of non - lte .
a summary of the abundance ratios for our m71 sample is shown in figure [ summ_fig2 ] .
the results for each element are depicted as a box whose central horizontal line is the median abundance ratio for all the m71 stars included , while the bottom and the top shows its inter quartile range , the vertical lines coming out of the box mark the position of the adjacent points of the sample , and the outliers are plotted as open circles .
the boxes drawn with dotted lines correspond to elements with abundances computed from only one line in each star and hence are more uncertain .
the thick line on the left side of the box is the predicted 1@xmath27 rms error scaled to correspond to the @xmath1025% inter quartile range .
as seen in figure [ summ_fig2 ] , the elements where we expect to see star - to - star variations in our m71 sample are o , na , zr , and the special cases of li and c , each to be discussed in detail later .
the abundance ratios of [ sc / fe ] , [ v / fe ] , [ cr / fe ] , [ mn / fe ] , [ co / fe ] , and [ ni / fe ] follow the behavior of iron as expected , showing no significant trend with @xmath2 , and less scatter around the mean than the predicted error
. the mean abundance ratios of sc ( @xmath14[sc / fe]@[email protected] ) , v ( @xmath14[v / fe]@[email protected] ) , and ni ( @xmath14[ni / fe]@[email protected] ) are consistent with the earlier results of @xcite , who analyzed high resolution spectra of ten giant stars in m71 , obtaining @xmath14[sc / fe]@[email protected] , @xmath14[v / fe]@[email protected] , and @xmath14[ni / fe]@[email protected] .
our abundance ratios of the iron peak elements are also consistent with the results of @xcite .
the zn i line at 6362.3 is definitely present in the best of the spectra of the m71 giants , but is somewhat blended with the much stronger fe i line at 6362.9 .
in addition , the continuum there is depressed due to a broad auto - ionization ca i feature .
the rather high abundance of zn we deduce must thus be regarded as quite uncertain until a full spectral synthesis of this region becomes available .
we have detected lines of the neutron capture elements y , zr , ba , la and eu . @xcite and @xcite analyzed the solar system meteoritic abundances of neutron capture elements to yield accurate breakdowns into @xmath29- and @xmath30-process parts for each isotope , which have been summed into fractions for each element by @xcite . at [ fe / h ]
@xmath312.0 , as reviewed by @xcite , zr , ba , and la are neutron capture element synthesized through @xmath32process reactions that occur mainly in low mass asymptotic giant branch ( agb ) stars , while eu is exclusively an @xmath33process element .
the abundance ratios of the neutron capture elements , y , zr , ba , la , and eu , show no significant trend with @xmath2 , and less scatter around the mean than the predicted error , except for [ zr / fe ] . in figure
[ zr_spec ] , we show the spectra for two stars of similar @xmath2 and different [ zr / fe ] , 156 ( 4525 k , [ zr / fe]=0.57 ) and 181 ( 4550 k , [ zr / fe]=0.00 ) in the region of the zr line at 6143 , which is the strongest zr i line included in our study .
in the spectral range illustrated , there are also two fe i lines , one ba ii line and one si i line , whose strengths are similar in both stars
. it is possible but not certain that the difference in strength of the zr line is due to star - to - star abundance variations .
@xcite analyzed four zr lines in five bright giant m71 stars to obtain [ zr / fe ] @xmath34 dex .
the abundances of ba , la , and eu are overabundant relative to fe , as is seen in other clusters ( see below ) .
the mean [ ba / eu ] ratio of + 0.03 is consistent with values observed in halo stars of similar [ fe / h ] @xcite .
we find that the @xmath1elements mg , ca , si and ti are overabundant relative to fe . our mean @xmath14[ti / fe]@[email protected] and @xmath14[si / fe]@[email protected] are similar to the results of @xcite for ten m71 giant stars ( @xmath14[ti / fe]@[email protected] , @xmath14[si / fe]@[email protected] ) , and also similar to the abundance ratios provided by @xcite . our @xmath14[ca / fe]@[email protected] is higher than the value of @xcite ( @xmath14[ca / fe]@[email protected] ) , but similar to the abundance ratio of + 0.58 found by @xcite .
the @xmath1element abundance ratios show no significant trend with @xmath2 , and low scatter around the mean .
[ mg / fe ] is know to vary among bright giant stars in some metal poor globular clusters . in ngc
6752 @xcite , m13 @xcite and m15 @xcite , [ mg / fe ] shows a star - to - star range in abundance of about 1.5 , 1.2 , and 1.0 dex respectively .
our comparison between the observed scatter and the predicted error of [ mg / fe ] given in table [ tab7 ] indicates no sign of star - to - star variation of magnesium in m71 .
the oxygen abundance ratios in our sample of stars in m71 behave differently than the abundance ratios of all other elements included in this paper .
the scatter in [ o / fe ] versus @xmath2 shown in figure [ light ] strongly suggests that o shows star - to - star variations within the m71 sample .
furthermore , the observed scatter for [ o / fe ] given in table [ tab7 ] is larger than the respective predicted error which include the effects of uncertainties in the determination of the the stellar parameters and in the equivalent width measurements . to a lesser extent ,
the na abundances behave similarly , as shown in figure [ light ] , and the observed scatter for na is slightly larger than the value @xmath20 given in table [ tab7 ] . in figure [ na_o_spec ]
we compare the strength of the na i and o i lines between two stars of similar stellar parameters .
the star with a high [ o / fe ] in our sample ( 160 ) has a low [ na / fe ] and the star with a low [ o / fe ] ( 2160 ) has an intermediate [ na / fe ] .
these two stars are marked with open squares in figure [ light ] .
note that both of these stars are red giants fainter than the hb , with @xmath35 for the fainter , na - richer star .
this figure demonstrates that the higher scatter seen in [ na / fe ] and [ o / fe ] is due to star - to - star abundance variations and that in the case of this specific pair of stars , na and o are anti - correlated , as was first observed by @xcite in m13 .
the non - lte corrections for the infrared oi triplet are not accurately know ( see sec .
[ abun ] ) .
however , assuming that they are monotonically dependent on @xmath2 , the rms in the non - lte correction can not introduce a star - to - star scatter in the o abundance nor the na - o anticorrelation we observe . to explore the presence of an anti - correlation between [ na / fe ] and [ o / fe ] within our m71 sample as a whole , we construct figure [ na_o_m71 ] , which presents the na versus o abundance diagram for our sample in m71 .
our data are indicated by filled symbols , where triangles are red giants brighter than the hb , circles are hb stars , squares are red giants fainter than the hb and the stars near the main sequence turnoff are denoted by `` stars '' .
we find that na and o are anti - correlated in our sample of m71 stars .
the fit weighted by the error of na vs. o , plotted as a solid line in figure [ na_o_m71 ] , is statistically significant at a 2@xmath27 confidence level .
star - to - star na variations and anti - correlation between na and o abundances extend well beyond @xmath35 , and include the small sample of stars near the main sequence turnoff , where this anti - correlation has the same form as among the more luminous stars and is highly statistically significant at a level exceeding 4@xmath27 . in view of the large sample of bright rgb stars studied in m4 by @xcite
, we adopt their results for the observed anti - correlation between na and o among red giants in this cluster to provide a fiducial line for visual comparisons in the relevant figures .
the anti - correlation found from their sample is indicated as a dashed line in figure [ na_o_m71 ] , as well as in the panels of figure [ na_o ] , to be discussed next . to within the errors ,
the na / o anti - correlation we find in m71 agrees with that of m4 , within a 2@xmath27 confidence level . in figure [ na_o ]
, we compare the determinations of na and o abundance ratios that exist in the literature for metal poor globular clusters , 47 tuc @xcite , m71 ( * ? ? ?
* , this paper ) , m5 @xcite , m4 @xcite , ngc 6752 @xcite , m3 @xcite , m10 @xcite , m13 @xcite , ngc 6397 @xcite , m92 @xcite , and m15 @xcite .
the symbols are the same as in figure [ na_o_m71 ] .
also included in this figure are the earlier results for ten bright giant stars in m71 from @xcite , shown in open triangles .
all their stars are brighter than the hb and behave similarly to our red giants brighter than the hb .
our observed range in [ na / fe ] is similar to the range observed by @xcite , but our range in [ o / fe ] is twice as big . for each globular cluster depicted in figure [ na_o ] ,
the solid line represent the least squares linear fit of the data from the literature .
it is only shown for those globular clusters where the slope we derive is significant at the 2@xmath27 level .
the dashed line corresponds to the anti - correlation observed in m4 from @xcite , shown as a fiducial line . at this confidence level
, we find na - o anti - correlations in m71 , m5 , m4 , ngc 6752 , m3 , m10 , m13 , m92 and m15 .
the steepest slope is that of m92 , and the flattest slope is that of m13 .
but , within the 2@xmath27 level , all the slopes are identical .
no statistically significant global anti - correlation is detected in 47 tuc or ngc 6397 .
47 tuc ( [ fe / h]@xmath360.8 ) , m4 ( [ fe / h]@xmath361.2 ) and ngc 6397 ( [ fe / h]@xmath362.0 ) have a similar [ na / fe ] versus [ o / fe ] relationship as does m71 does in terms of abundance ratio ranges and scatter .
while the form of the relationship appears to be more or less universal , figure [ na_o ] suggests that the amplitude of the na / o anti - correlation among rgb stars is smallest for the two most metal rich globular clusters shown , 47 tuc ( where the published dataset is very small ) and m71 , as well as for ngc 6397 .
the abundance of al is also known to vary from star - to - star within a globular cluster . because the al doublet at 6697 is not included in the spectral coverage of the primary set of hires spectra ( see paper i ) , it can be measured only in a subset of the sample of stars studied here .
our comparison between @xmath19 and the predicted error of [ al / fe ] given in table [ tab7 ] indicates that the scatter for this element abundance ratio is slightly larger to its respective predicted error . as discussed in
[ section_nonlte ] , al suffers from non - lte effects .
we have not adopted any corrections , nor have we applied any to the set of data from the literature assembled for al .
we do , however , use the 6696 doublet , which is less susceptible to non - lte effects than is the 3961 line .
we have constructed figure [ al_na_m71 ] to explore the presence of a correlation between [ na / fe ] and [ al / fe ] in m71 , seen in other globular clusters .
the symbols in figure [ al_na_m71 ] are the same as in figure [ na_o_m71 ] .
we include only our own data in this figure .
a clear al / na correlation is seen which is statistically significant at a 2@xmath27 level .
this correlation extends down to @xmath3 mag in m71 , where the sample ends due to the technical issue of the hires spectral coverage . in figure [ al_na ]
, we compare the determinations of al / na abundance ratios that exist in the literature for metal poor globular clusters .
we use the same set of references as in [ section_o ] , although there are fewer stars with measured al abundances .
the symbols are the same as in figure [ na_o_m71 ] . again the solid line , representing the least squares linear fit of the data , is only shown in those globular clusters where the slope we derive is significant . at the 2@xmath27 level , we find na - al correlations in m5 , m4 , ngc 6752 , m13 , ngc 6397 , m92 and m15 , as well as in m71 . at this confidence level , all the slopes , with the exception of that of m13 , are identical .
no anti - correlation is detected in 47 tuc , where the database is very sparse .
the differences among the family of linear fits to the al / na relationship for various globular clusters shown in figure [ al_na ] appear at first sight to be considerably larger than those shown by the fits to the na / o anti - correlation in figure [ na_o ] .
we suggest that these differences may not be real , and may arise from non - lte effects in al acting on the different ranges of luminosities of stars studied in each cluster as well as the particular selection of al lines used in each analysis .
the situation in ngc 6752 is particularly illuminating .
et al . _ ( 2001 ) ascribe the very different mean al abundances deduced for the subgiants and for the main sequence stars in the cluster precisely to this issue of ignoring non - lte in the spectrum of aluminium .
li is a very fragile element and is very easily destroyed in stars , burning at @xmath37 k. @xcite discovered the presence of li in warm halo dwarfs at a constant value ( log @xmath16(li ) = 2.24 ) and suggested that this represents the primordial li synthesized in the big bang , thus of considerable importance to cosmology .
destruction of li is a measure of the depth of the surface convection zone , and hence a strong function of @xmath2 .
@xcite compile recent observations for li in galactic disk and halo stars and review the galactic chemical evolution of li , while @xcite reviews the destruction of li from a theoretical perspective .
we therefore expect li to be depleted among the rgb and subgiant stars in globular clusters , although probably not among the main sequence stars .
in addition , there is at least one case known in a globular cluster of the extremely rare class of very li - rich stars .
a possible explanation for this star , found as a bright rgb star in m3 by @xcite , and similar objects is given by @xcite .
the li line is not included in the spectral coverage of the primary set of spectra ( see paper i ) , and hence can be measured only in a subset of the sample of m71 stars studied here .
we were able to obtain log @xmath16(li ) for three giants fainter than the hb , as well as several upper limits .
the mean log @xmath16(li ) for the detections is [email protected] ( on the scale h=12.0 dex ) , which is 0.8 dex less than the mean log @xmath16(li ) ( 1.90 @xmath10 0.42 ) for a sample of 11 halo dwarf stars of similar [ fe / h ] from the sample of @xcite , and is evidence of the strong depletion among the cooler stars in which li was detected here .
figure [ li1 ] illustrates the pattern of detections and upper limits , which are consistent with our overall expectations .
the analysis of c i lines in cool stars is difficult as the lines are weak , and their excitation potential is high , @xmath368.5 ev .
furthermore , the c i lines near 7115 are not included in the spectral coverage of the primary set of spectra ( see paper i ) , and hence can be measured only in a subset of the sample studied here .
we have reliable detections in only 6 stars , 2 lines in 1 star , and 1 line each in the other 5 stars .
these , with considerable uncertainty , show a large star - to - star scatter in deduced c abundance .
however , the c i lines in stars with @xmath2 @xmath38 k may be blended with or completely dominated by lines from the red system of cn , as illustrated in the spectrum of arcturus by @xcite .
it is very likely that this has happened , as there are two cool stars in our sample , m71 1 - 45 and 1 - 66 , with anomalously high deduced c abundances .
the measurements of @xcite show that these two stars have much stronger cn lines than does m71 star i , a star of similar @xmath2 which yielded a much more reasonable c abundance .
unpublished measurements of the g band of ch in these three stars by jgc also suggest that the very large c abundances we deduce for m71 1 - 45 and 1 - 66 are spurious .
the molecular band data gives a much clearer picture of the pattern of c abundance variation in m71 as the samples are much larger and the c abundance can be inferred with considerable precision from the strength of the ch band .
both the ch and cn bands clearly show strong star - to - star variations on the m71 giant branch ( * ? ? ?
* and references therein ) and , more importantly , at the level of the main sequence @xcite .
the entire set of molecular band data can be explained by a variation in c of about a factor of 2 , with a much larger anti - correlated variation in the n abundance . in figures
[ new_comp1 ] to [ new_comp4 ] , for each element studied here we provide a comparison to similar high - resolution abundance analyses of halo dwarfs , of rgb stars in m4 ( [ fe / h]@xmath361.2 ) , rgb stars in m5 ( [ fe / h]@xmath361.2 ) , and rgb stars in m15 from @xcite , @xcite , @xcite , and @xcite , respectively .
the halo dwarfs plotted in the figures have been selected from the sample of @xcite to have [ fe / h ] similar to m71 ( @xmath39[fe / h]@xmath40 ) .
the boxes in figures [ new_comp1 ] to [ new_comp4 ] follow the same layout as those in figure [ summ_fig2 ] . the globular cluster name is indicated above each box ( h stands for the halo dwarf sample ) , and the number in parenthesis below the name indicates the number of stars used in the calculation of the respective abundance ratio .
the median abundances for each element determined in each of the five different environments presented in figures [ new_comp1 ] to [ new_comp4 ] agree to within the 1@xmath27 uncertainties of each measurement for most of the elements displayed , and agree to within @xmath41 for _ all _ the elements shown .
aluminium is the element showing the largest trend with metallicity between these five environments .
this is perhaps a consequence of not considering non - lte effects and of some studies including the 3961 doublet , known to be more sensitive to non - lte effects , and others not .
barium is the only other element showing large variations in its median abundance among the various environments , although no consistent trend with metallicity .
we suggest that this too may not be real , and may be a reflection of the issue of hyperfine structure corrections in the fairly strong lines of this element .
overall m71 appears to have very similar abundance ratios as does m5 . in this set of figures ,
one can see some of the well known trends characteristic of halo star abundances as reviewed by @xcite , such as the gradual increase of [ @xmath42/fe ] as [ fe / h ] decreases , particularly for si and ti .
a classical review of post - main sequence stellar evolution can be found in @xcite .
their description of the consequences of the first dredge up phase , the only dredge up phase any of the stars in our m71 sample may have experienced , indicates that a doubling of the surface n@xmath43 and a 30% reduction in the surface c@xmath44 can be expected , together with a drop in the ratio of c@xmath44/c@xmath45 from the solar value of 89 to @xmath3620 , as well as a drop in surface li and b by several orders of magnitude .
observations of field stars over a wide range of luminosities conform fairly well to this picture , see e.g. @xcite .
however , the o / na anti - correlation seen among the bright red giants in many globular clusters , including here in the case of m71 , can not be explained in this picture .
several theoretical mechanisms have thus been proposed ( e.g. , the meridional mixing of sweigart & mengel 1979 , and turbulent diffusion of charbonnel 1994 , 1995 ) with varying degrees of success .
in addition , @xcite suggested that the nuclear reaction @xmath46ne(@xmath47na occurs in regions of the h - burning shell for low mass stars where o is converted into n and produces na@xmath48 and al@xmath49 . @xcite
combined these ideas to predict the consequences of such possible synthesis and deep mixing , including for example , that the surface mg abundance should be much less affected than that of na or al .
these ideas form the basis of our current understanding of dredge up in low mass metal poor giants , with more recent calculations given by @xcite , among others .
the clear prediction of this suite of calculations is that the earliest that deep mixing can begin is at the location of the bump in the luminosity function of the rgb which occurs when the h - burning shell crosses a sharp chemical discontinuity . @xcite
have shown that the luminosity of the rgb bump as a function of metallicity as determined from observation agrees well with that predicted by the theory of stellar evolution .
@xcite further suggest that the agreement between the predicted luminosity function and actual star counts along the rgb in the vicinity of the bump in a suite of globular clusters is so good that mixing ( more exactly , mixing of he ) can not have occurred any earlier , otherwise the evolutionary lifetimes , and hence the observed lf , of such stars would have been affected .
our sample of m71 stars shows a statistically significant correlation between al and na abundances which extends to stars as faint as @xmath3 mag .
we also see an anti - correlation between the na and o abundances extending down to near the main sequence turn off .
we see variations in li ( as expected ) , and may see variations in zr ( not expected ) .
any variations in mg are smaller than those of al , na or o ( as expected ) .
we know there are large anti - correlated c and n variations from the work of @xcite ; this too is expected .
the behavior of li , which is very fragile and easily destroyed , is not controversial .
it is , however , the range of luminosity over which the remainder of these variations are seen which is becoming more and more of a problem for any scenario which invokes dredge up and mixing .
the na / o anti - correlation we see in m71 extends to the main sequence turn off .
the al / na correlation we see in m71 extends to at least as faint as @xmath3 , while the rgb bump in a cluster of the metallicity of m71 is at @xmath50 .
@xcite has shown that the c / n anti - correlation extends to the stars at the main sequence turnoff and even fainter in m71 . @xcite
have shown that the c / n abundance range seen at the level of the main sequence is comparable to that seen among the bright red giants of m71 by many previous studies , the most recent of which is @xcite . the accumulated weight of recent evidence , both in m71 as described above and in other globular clusters such as 47 tuc ( see * ? ? ? * and references therein ) and ngc 6752 ( see * ? ? ?
* and references therein ) , suggests that we are now back in the situation we were during the late 1980s .
unless we have missed some important aspect of stellar evolution with impact on mixing and dredge up , we must declare the mixing scenario a failure for the specific case of m71 ( and several other globular clusters as well ) .
even the theoreticians in the forefront of this field are beginning to admit that deep mixing alone is not sufficient @xcite .
unless and until some major new concept relevant to this issue appears , we must now regard the fundamental origin of the star - to - star variations we see in m71 as arising outside the stars whose spectra we have studied here .
we present results of a high dispersion analysis of 23 elements to obtain abundance ratios for 25 members of the galactic globular cluster m71 .
our sample of stars includes 19 giant stars ( 9 of which are less luminous than the rhb ) , 3 horizontal branch stars , and 3 stars near the main sequence turnoff .
our conclusions are summarized as follows : * the iron peak and neutron capture element abundance ratios show no trend with @xmath2 , and low scatter around the mean . * the @xmath1elements mg , ca , si and ti are overabundant relative to fe . the scatter about
the mean is small . *
an anti - correlation between the na and o abundances for stars in m71 is detected with a statistical significance in excess of 2@xmath27 , and extends to the stars near the main sequence turn off . *
the [ na / fe ] versus [ al / fe ] correlation is detected with a statistical significance in excess of 2@xmath27 in our sample of m71 stars , and extends at least as faint as @xmath3 . * both extremely o poor , na rich stars and extremely o rich , na poor stars such as are observed in m5 and in m13 are not present in our sample of m71 stars .
* li is varying among the subgiants ( as expected ) , and zr may be varying among the subgiants .
m71 appears very similar in its element abundance ratios to m5 , which is not surprising as m5 has a metallicity only slightly lower , [ fe / h ] = @xmath51 dex @xcite .
however , the amplitude of the na / o and al / na relationships appears to be somewhat larger in m5 than in m71 , and still larger in even more metal poor clusters .
our detailed abundance analysis of 25 stars in m71 has revealed abundance variations appearing at such low luminosities that deep mixing scenarios can no longer reproduce these results .
this problem is made even more acute when we add in the data of @xcite and the analysis of @xcite of the ch and cn bands in m71 .
we are forced to the firm conclusion that much , if not all , of the abundance variations seen in m71 must have been in place before the present generation of stars we observe was formed , or ( less likely ) are the result of some type ( binaries ? ) of mass transfer . in future papers
, we will proceed to apply the techniques and analysis developed here for m71 to other more metal poor globular clusters , where the rgb bump is predicted to be somewhat more luminous and where , judging from the behavior of their bright rgb stars , we may anticipate finding even larger variations at low luminosities among the cluster subgiants and main sequence stars .
the entire keck / hires user communities owes a huge debt to jerry nelson , gerry smith , steve vogt , and many other people who have worked to make the keck telescope and hires a reality and to operate and maintain the keck observatory .
we are grateful to the w. m. keck foundation for the vision to fund the construction of the w. m. keck observatory .
the authors wish to extend special thanks to those of hawaiian ancestry on whose sacred mountain we are privileged to be guests . without their generous hospitality , none of the observations presented herein would have been possible .
we are grateful to the national science foundation for partial support under grant ast-9819614 to jgc .
we thank jason prochaska and andy mcwilliam for providing their tables of hyperfine structure in digital form .
gratton , r. g. , bonifacio , p. , bragaglia , a. , carretta , e. , castellani , v. , centurion , m. , chieffi , a. , claudi , r. , clementini , g. , dantona , f. , desidera , s. , francois , p. , grundahl , f. , lucatello , s. , molaro , p. , pasquini , l. , sneden , c. , spite , f. , & straniero , o. , 2001 , , 369 , 87 lcccrcrc 145 & 3950 & 0.90 & 1.45 & 187 & 0.60@xmath10 0.03 & 5 & 0.45@xmath10 0.06 + i & 4150 & 1.00 & 1.38 & 186 & 0.63@xmath10 0.03 & 6 & 0.66@xmath10 0.07 + 166 & 4250 & 1.35 & 1.35 & 179 & 0.58@xmath10 0.03 & 6 & 0.67@xmath10 0.09 + 164 & 4200 & 1.35 & 1.37 & 187 & 0.61@xmath10 0.03 & 5 & 0.52@xmath10 0.09 + 156 & 4525 & 1.60 & 1.26 & 127 & 0.48@xmath10 0.03 & 2 & 0.68@xmath10 0.13 + 195 & 4550 & 1.65 & 1.25 & 184 & 0.59@xmath10 0.03 & 8 & 0.74@xmath10 0.05 + 181 & 4550 & 1.75 & 1.25 & 180 & 0.56@xmath10 0.03 & 6 & 0.93@xmath10 0.05 + 11 & 4700 & 2.05 & 1.20 & 134 & 0.59@xmath10 0.03 & 5 & 0.76@xmath10 0.05 + 180 & 5300 & 2.40 & 1.80 & 71 & 0.64@xmath10 0.03 & 5 & 0.93@xmath10 0.05 + 187 & 5300 & 2.40 & 1.80 & 128 & 0.56@xmath10 0.03 & 9 & 0.84@xmath10 0.05 + 194 & 5300 & 2.40 & 1.80 & 94 & 0.73@xmath10 0.03 & 6 & 0.82@xmath10 0.05 + 160 & 4900 & 2.30 & 1.13 & 119 & 0.74@xmath10 0.03 & 6 & 0.62@xmath10 0.05 + 159 & 4600 & 2.30 & 1.23 & 141 & 0.70@xmath10 0.03 & 5 & 0.58@xmath10 0.05 + g53476_4543 & 4900 & 2.65 & 1.13 & 174 & 0.61@xmath10 0.03 & 7 & 0.76@xmath10 0.05 + 2160 & 5100 & 2.70 & 1.07 & 145 & 0.49@xmath10 0.03 & 5 & 0.94@xmath10 0.08 + g53447_4707 & 5175 & 2.75 & 1.04 & 155 & 0.51@xmath10 0.03 & 7 & 0.83@xmath10 0.05 + g53445_4647 & 5050 & 2.85 & 1.08 & 112 & 0.59@xmath10 0.03 & 6 & 0.80@xmath10 0.05 + g53447_4703 & 5000 & 3.00 & 1.10 & 125 & 0.71@xmath10 0.03 & 4 & 0.74@xmath10 0.05 + g53425_4612 & 5150 & 3.15 & 1.05 & 80 & 0.67@xmath10 0.03 & 2 & 0.87@xmath10 0.07 + g53477_4539 & 5150 & 3.15 & 1.05 & 119 & 0.64@xmath10 0.03 & 5 & 0.86@xmath10 0.05 + g53457_4709 & 5200 & 3.35 & 1.03 & 93 & 0.72@xmath10 0.03 & 5 & 0.72@xmath10 0.12 + g53391_4628 & 5100 & 3.35 & 1.07 & 106 & 0.78@xmath10 0.03 & 5 & 0.76@xmath10 0.05 + g53417_4431 & 5800 & 4.05 & 0.83 & 38 & 0.63@xmath10 0.04 & 3 & 0.60@xmath10 0.12 + g53392_4624 & 5800 & 4.05 & 0.83 & 36 & 0.76@xmath10 0.03 & 3 & 0.66@xmath10 0.08 + g53414_4435 & 5900 & 4.15 & 0.80 & 13 & 0.78@xmath10 0.05 & 2 & 0.58@xmath10 0.16 + llrrcccccccccccc c i & 7113.180 & 8.640 & -0.773 & 35.5 & 16.1 & 38.4 & ... & ... & 30.0 & ... & ... & ... & ... & ... & ... + c i & 7115.170 & 8.640 & -0.710 & 31.3 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... + o i & 6300.304 & 0.000 & -9.780 & 76.2 & 73.8 & 57.1 & 71.6 & 42.4 & 44.9 & 44.5 & 30.5 & 18.8 & 23.4 & 20.1 & 30.4 + o i & 6363.776 & 0.020 & -10.300 & 39.3 & 34.3 & 25.9 & 35.7 & 19.2 & 19.6 & 22.3 & ... & ... & ... & 9.0 & 14.9 + o i & 7771.944 & 9.150 & 0.369 & ... & ... & 15.2 & 21.3 & 23.2 & 21.2 & 19.9 & 28.7 & 65.5 & 52.5 & 81.0 & 44.4 + o i & 7774.166 & 9.150 & 0.223 & ... & 17.2 & 16.1 & 18.8 & 17.5 & 23.8 & 17.5 & 21.4 & 63.6 & 50.9 & 62.3 & 43.6 + o i & 7775.388 & 9.150 & 0.001 & ... & 18.6 & ... & 14.3 & 18.7 & 14.7 & 14.7 & 14.3 & 32.3 & 31.3 & 50.6 & 32.5 + na i & 5682.633 & 2.100 & -0.700 & 159.9 & 130.6 & 137.0 & 123.8 & 121.5 & 105.0 & 106.8 & 106.4 & 66.1 & 68.6 & 63.1 & 64.4 + na i & 5688.193 & 2.100 & -0.420 & 159.5 & 134.3 & 146.3 & 141.0 & 131.5 & 125.5 & 129.3 & 125.3 & 84.4 & 90.4 & 83.4 & 84.9 + na i & 6154.225 & 2.100 & -1.530 & 108.2 & 60.4 & 72.3 & 64.5 & 52.8 & 43.0 & 47.3 & 48.4 & 7.0 & 15.7 & 13.4 & 18.4 + na i & 6160.747 & 2.100 & -1.230 & 136.6 & 91.1 & 101.3 & 93.3 & 89.5 & 64.3 & 71.2 & 68.6 & 27.6 & 32.0 & 17.3 & 30.6 + mg i & 5711.088 & 4.340 & -1.670 & 137.7 & 139.4 & 136.0 & 126.3 & 128.1 & 126.3 & 124.1 & 103.9 & 100.1 & 102.2 & 97.9 & 107.3 + mg i & 6318.717 & 5.110 & -1.970 & 69.0 & 63.8 & 55.9 & 56.7 & 64.2 & 53.3 & 54.9 & 61.2 & ... & 37.9
& ... & 37.4 + mg i & 6319.237 & 5.110 & -2.220 & 55.9 & 51.0 & 50.2 & 53.9 & ... & 38.7 & 43.1 & ...
& ... & 26.9 & 25.3 & ... + mg i & 6319.495 & 5.110 & -2.680 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... + mg i & 6965.409 & 5.750 & -1.870 & 35.7 & 34.9 & 41.8 & 40.8 & ... & 37.6 & 42.5 & 31.3 & ... & ... & ... & ... + mg i & 7193.184 & 5.750 & -1.400 & 68.4 & 61.4 & 71.1 & 63.7 & 55.5 & 53.5 & 52.9 & ... & ... & 41.0 & ... & 37.3 + mg i & 7387.689 & 5.750 & -0.870 & 51.3 & 66.9 & 62.2 & 54.4 & ... & 52.9 & 61.4 & 48.8 & ... & 51.6 & 41.6 & ... + mg i & 7657.603 & 5.110 & -1.280 & 112.5 & 112.5 & 111.6 & 112.5 & 97.3 & 98.0 & 101.9 & 91.4 & 77.4 & 77.4 & ... & ... + mg i & 7722.601 & 5.940 & -1.800 & ... & ... & 35.6 & ... & ... & 32.8 & 46.8 & ... & ... & ... & ... & ... + mg i & 7930.810 & 5.940 & -1.200 & 53.6 & 60.2 & 67.9 & 56.5 & ... & 62.2 & 52.6 & 45.0 & ... & ... & ... & ... + al i & 5557.059 & 3.140 & -2.100 & ... & ... & 29.2 & 22.0 & ... & 14.1 & 18.8 & ... & ... & ... & ... & ... lcccc li i & 1 & 0.94 & ... & 0.22 + c i & 4 & + 1.08 & 0.04 & + 0.03 + o i & 5 & + 1.53 & 0.10 & + 0.11 + na i & 4 & 1.30 & 0.09 & 0.10 + mg i & 10 & + 0.03 & 0.24 & 0.05 + al i & 6 & 1.20 & 0.15 & 0.17 + si i & 20 & + 0.14 & 0.12 & + 0.09 + k i & 1 & 2.28 & ... & + 0.10 + ca i & 15 & 1.56 & 0.14 & 0.39 + sc ii & 7 & 4.26 & 0.12 & + 0.16 + ti i & 40 & 2.48 & 0.15 & + 0.09 + v i & 13 & 3.55 & 0.14 & 0.06 + cr i & 12 & 1.72 & 0.16 & + 0.10 + mn i & 4 & 2.09 & 0.12 & 0.12 + co i & 7 & 2.60 & 0.08 & 0.00 + ni i & 43 & 1.19 & 0.18 & + 0.07 + cu i & 1 & 3.44 & ... & 0.20 + zn i & 1 & 2.94 & ... & 0.09 + y ii & 1 & 4.96 & ... & + 0.33 + zr i & 4 & 4.52 & 0.16 & + 0.38 + ba ii & 3 & 5.29 & 0.08 & + 0.01 + la ii & 3 & 6.27 & 0.07 & + 0.04 + eu ii & 1 & 6.96 & ... & + 0.01 + lrrrrrrrr 145 & 0 & ... & 2 & [email protected] & 2 & [email protected] & 4 & [email protected] + i & 0
& ... & 1 & [email protected] & 4 & [email protected] & 4 & [email protected] + 166 & 1 & @xmath14 0.61 & 1 & [email protected] & 4 & [email protected] & 4 & [email protected] + 164 & 1 & @xmath14 0.79 & 0 & ... & 5 & [email protected] & 4 & [email protected] + 156 & 0 & ... & 0 & ... & 5 & [email protected] & 4 & [email protected] + 195 & 1 & @xmath140.39 & 1 & [email protected] & 5 & [email protected] & 4 & [email protected] + 181 & 0 & ... & 0 & ... & 5 & [email protected] & 4 & [email protected] + 11 & 1 & @xmath14 0.36 & 0 & ... & 4 & [email protected] & 4 & [email protected] + 180 & 1 & @xmath140.69 & 0 & ... & 4 & [email protected] & 4 & [email protected] + 187 & 0 & ... & 0 & ... & 4 & [email protected] & 4 & [email protected] + 194 & 1 & @xmath140.69 & 0 & ... & 5 & [email protected] & 4 & [email protected] + 160 & 0 & ... & 0 & ... & 5 & [email protected] & 4 & [email protected] + 159 & 0 & ... & 0 & ... & 5 & [email protected] & 4 & [email protected] + g53476_4543 & 0 & ... & 0 & ... & 5 & [email protected] & 4 & [email protected] + 2160 & 1 & @xmath14 0.61 & 1 & [email protected] & 4 & [email protected] & 3 & [email protected] + g53447_4707 & 1 & [email protected] & 0 & ... & 4 & [email protected] & 4 & [email protected] + g53445_4647 & 1 & [email protected] & 1 & [email protected] & 4 & [email protected] & 4 & [email protected] + g53447_4703 & 0 & ... & 0 & ... & 3 & [email protected] & 4 & [email protected] + g53425_4612 & 1 & [email protected] & 0 & ... & 3 & [email protected] & 4 & [email protected] + g53477_4539 & 0 & ... & 0 & ... & 4 & [email protected] & 4 & [email protected] + g53457_4709 & 0 & ... & 0 & ... & 3 & [email protected] & 4 & [email protected] + g53391_4628 & 0 & ... & 0 & ... & 3 & [email protected] & 4 & [email protected] + g53417_4431 & 0 & ... & 0 & ... & 3 & [email protected] & 4 & [email protected] + g53392_4624 & 0 & ... & 0 & ... & 3 & [email protected] & 3 & [email protected] + g53414_4435 & 0 & ... & 0 & ... & 3 & [email protected] & 3 & [email protected] lrrrrrrrrrr 145 & 8 & [email protected] & 3 & [email protected] & 12 & [email protected] & 1 & [email protected] & 15 & [email protected] + i & 8 & [email protected] & 3 & [email protected] & 14 & [email protected] & 1 & [email protected] & 13 & [email protected] + 166 & 9 & [email protected] & 3 & [email protected] & 17 & [email protected] & 1 & [email protected] & 14 & [email protected] + 164 & 8 & [email protected] & 4 & [email protected] & 16 & [email protected] & 1 & [email protected] & 15 & [email protected] + 156 & 4 & [email protected] & 0 & ... & 10 & [email protected] & 0 & ... & 10 & [email protected] + 195 & 9 & [email protected] & 3 & [email protected] & 17 & [email protected] & 1 & [email protected] & 13 & [email protected] + 181 & 9 & [email protected] & 4 & [email protected] & 16 & [email protected] & 1 & [email protected] & 14 & [email protected] + 11 & 6 & [email protected] & 1 & [email protected] & 14 & [email protected] & 1 & [email protected] & 12 & [email protected] + 180 & 2 & [email protected] & 0 & ... & 10 & [email protected] & 0 & ... & 10 & [email protected] + 187 & 6 & [email protected] & 0 & ... & 16 & [email protected] & 0 & ... & 12 & [email protected] + 194 & 3 & [email protected] & 2 & [email protected] & 13 & [email protected] & 1 & [email protected] & 10 & [email protected] + 160 & 3 & [email protected] & 0 & ... & 12 & [email protected] & 0 & ... & 11 & [email protected] + 159 & 4 & [email protected] & 0 & ... & 16 & [email protected] & 0 & ... & 12 & [email protected] + g53476_4543 & 5 & [email protected] & 0 & ... & 18 & [email protected] & 0 & ... & 15 & [email protected] + 2160 & 7 & [email protected] & 2 & [email protected] & 17 & [email protected] & 0 & ... & 13 & [email protected] + g53447_4707 & 7 & [email protected] & 2 & [email protected] & 17 & [email protected] & 1 & [email protected] & 12 & [email protected] + g53445_4647 & 4 & [email protected] & 1 & [email protected] & 13 & [email protected] & 0 & ... & 12 & [email protected] + g53447_4703 & 3 & [email protected] & 0 & ... & 16 & [email protected] & 1 & [email protected] & 13 & [email protected] + g53425_4612 & 5 & [email protected] & 0 & ... & 4 & [email protected] & 1 & [email protected] & 9 & [email protected] + g53477_4539 & 6 & [email protected] & 0 & ... & 17 & [email protected] & 0 & ... & 11 & [email protected] + g53457_4709 & 4 & [email protected] & 0 & ... & 11 & [email protected] & 0 & ... & 11 & [email protected] + g53391_4628 & 2 & [email protected] & 0 & ... & 15 & [email protected] & 0 & ... & 10 & [email protected] + g53417_4431 & 1 & [email protected] & 0 & ... & 5 & [email protected] & 0 & ... & 11 & [email protected] + g53392_4624 & 1 & [email protected] & 0 & ... & 4 & [email protected] & 0 & ... & 8 & [email protected] + g53414_4435 & 0 & ... & 0 & ... & 0 & ... & 0 & ... & 8 & [email protected] lrrrrrrrr 145 & 6 & [email protected] & 36 & [email protected] & 14 & [email protected] & 10 & [email protected] + i & 7 & [email protected] & 37 & [email protected] & 14 & [email protected] & 9 & [email protected] + 166 & 6 & [email protected] & 38 & [email protected] & 13 & [email protected] & 9 & [email protected] + 164 & 7 & [email protected] & 37 & [email protected] & 14 & [email protected] & 8 & [email protected] + 156 & 4 & [email protected] & 21 & [email protected] & 9 & [email protected] & 3 & [email protected] + 195 & 6 & [email protected] & 36 & [email protected] & 14 & [email protected] & 9 & [email protected] + 181 & 7 & [email protected] & 34 & [email protected] & 14 & [email protected] & 9 & [email protected] + 11 & 7 & [email protected] & 25 & [email protected] & 14 & [email protected] & 5 & [email protected] + 180 & 4 & [email protected] & 8 & [email protected] & 1 & [email protected] & 1 & [email protected] + 187 & 7 & [email protected] & 16 & [email protected] & 10 & [email protected] & 4 & [email protected] + 194 & 6 & [email protected] & 5 & [email protected] & 4 & [email protected] & 3 & [email protected] + 160 & 5 & [email protected] & 10 & [email protected] & 3 & [email protected] & 3 & [email protected] + 159 & 5 & [email protected] & 23 & [email protected] & 11 & [email protected] & 7 & [email protected] + g53476_4543 & 6 & [email protected] & 32 & [email protected] & 11 & [email protected] & 9 & [email protected] + 2160 & 6 & [email protected] & 24 & [email protected] & 12 & [email protected] & 7 & [email protected] + g53447_4707 & 7 & [email protected] & 21 & [email protected] & 9 & [email protected] & 8 & [email protected] + g53445_4647 & 7 & [email protected] & 15 & [email protected] & 8 & [email protected] & 6 & [email protected] + g53447_4703 & 7 & [email protected] & 17 & [email protected] & 6 & [email protected] & 6 & [email protected] + g53425_4612 & 5 & [email protected] & 6 & [email protected] & 7 & [email protected] & 3 & [email protected] + g53477_4539 & 5 & [email protected] & 13 & [email protected] & 7 & [email protected] & 3 & [email protected] + g53457_4709 & 6 & [email protected] & 8 & [email protected] & 3 & [email protected] & 3 & [email protected] + g53391_4628 & 6 & [email protected] & 15 & [email protected] & 0 & ... & 3 & [email protected] + g53417_4431 & 3 & [email protected] & 6 & [email protected] & 0 & ... & 1 & [email protected] + g53392_4624 & 3 & [email protected] & 6 & [email protected] & 0 & ... & 0 & ... + g53414_4435 & 1 & [email protected] & 4 & [email protected] & 0 & ... & 0 & ... lrrrrrrrrrr 145 & 3 & [email protected] & 7 & [email protected] & 37 & [email protected] & 1 & [email protected] & 0 & ... + i & 4 & [email protected] & 7 & [email protected] & 35 & [email protected] & 1 & [email protected] & 0 & ... + 166 & 4 & [email protected] & 7 & [email protected] & 35 & [email protected] & 1 & [email protected] & 1 & [email protected] + 164 & 4 & [email protected] & 8 & [email protected] & 38 & [email protected] & 1 & [email protected] & 1 & [email protected] + 156 & 0 & ... & 3 & [email protected] & 29 & [email protected] & 0 & ... & 0 & ... + 195 & 4 & [email protected] & 7 & [email protected] & 37 & [email protected] & 1 & [email protected] & 0 & ... + 181 & 4 & [email protected] & 7 & [email protected] & 36 & [email protected] & 1 & [email protected] & 0 & ... + 11 & 4 & [email protected] & 6 & [email protected] & 29 & [email protected] & 1 & [email protected] & 0 & ... + 180 & 0 & ... & 0 & ... & 17 & [email protected] & 1 & [email protected] & 0 & ... + 187 & 4 & [email protected] & 4 & [email protected] & 27 & [email protected] & 1 & [email protected] & 1 & [email protected] + 194 & 3 & [email protected] & 0 & ... & 16 & [email protected] & 1 & [email protected] & 1 & [email protected] + 160 & 0
& ... & 2 & [email protected] & 19 & [email protected] & 1 & [email protected] & 0 & ... + 159 & 0 & ... & 3 & [email protected] & 28 & [email protected] & 1 & [email protected] & 0 & ... + g53476_4543 & 0 & ... & 3 & [email protected] & 36 & [email protected] & 1 & [email protected] & 0 & ... + 2160 & 3 & [email protected] & 7 & [email protected] & 28 & [email protected] & 1 & [email protected] & 1 & [email protected] + g53447_4707 & 3 & [email protected] & 4 & [email protected] & 34 & [email protected] & 1 & [email protected] & 1 & [email protected] + g53445_4647 & 0 & ... & 2 & [email protected] & 23 & [email protected] & 1 & [email protected] & 1 & [email protected] + g53447_4703 & 2 & [email protected] & 1 & [email protected] & 25 & [email protected] & 1 & [email protected] & 0 & ... + g53425_4612 & 2 & [email protected] & 0 & ... & 13 & [email protected] & 1 & [email protected] & 0 & ...
+ g53477_4539 & 0 & ... & 0 & ... & 24 & [email protected] & 1 & [email protected] & 0 & ... + g53457_4709 & 0 & ... & 0 & ... & 15 & [email protected] & 1 & [email protected] & 1 & [email protected] + g53391_4628 & 0 & ... & 2 & [email protected] & 20 & [email protected] & 1 & [email protected] & 0 & ... + g53417_4431 & 0 & ... & 0 & ... & 7 & [email protected] & 0 & ... & 0 & ... + g53392_4624 & 0 & ... & 0 & ... & 5 & [email protected] & 0 & ... & 0 & ... + g53414_4435 & 0 & ... & 0 & ... & 2 & [email protected] & 0 & ... & 0 & ... lrrrrrrrrrr 145 & 0 & ... & 3 & [email protected] & 3 & [email protected] & 2 & [email protected] & 1 & [email protected] + i & 1 & [email protected] & 4 & [email protected] & 3 & [email protected] & 1 & [email protected] & 1 & [email protected] + 166 & 0 & ... & 4 & [email protected] & 3 & [email protected] & 3 & [email protected] & 1 & [email protected] + 164 & 0 & ... & 3 & [email protected] & 2 & [email protected] & 2 & [email protected] & 1 & [email protected] + 156 & 0 & ... & 1 & [email protected] & 3 & [email protected] & 2 & [email protected] & 1 & [email protected] + 195 & 1 & [email protected] & 3 & [email protected] & 3 & [email protected] & 2 & [email protected] & 1 & [email protected] + 181 & 0 & ... & 2 & [email protected] & 3 & [email protected] & 2 & [email protected] & 1 & [email protected] + 11 & 0 & ... & 0 & ... & 2 & [email protected] & 1 & [email protected] & 0 & ... + 180 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 1 & [email protected] + 187 & 0 & ... & 1 & [email protected] & 3 & [email protected] & 0 & ... & 1 & [email protected] + 194 & 0 & ... & 0 & ... & 3 & [email protected] & 1 & [email protected] & 0 & ... + 160 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 0 & ... + 159 & 0 & ... & 0 & ... & 3 & [email protected] & 1 & [email protected] & 0 & ... + g53476_4543 & 0 & ... & 1 & [email protected] & 3 & [email protected] & 1 & [email protected] & 1 & [email protected] + 2160 & 0 & ... & 1 & [email protected] & 3 & [email protected] & 1 & [email protected] & 0 & ... + g53447_4707 & 1 & [email protected] & 0 & ... & 3 & [email protected] & 1 & [email protected] & 1 & [email protected] + g53445_4647 & 0 & ... & 0 & ... & 2 & [email protected] & 0 & ... & 0 & ... + g53447_4703 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 0 & ... + g53425_4612 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 0 & ... + g53477_4539 & 0 & ... & 0 & ... & 3 & [email protected] & 1 & [email protected] & 0 & ... + g53457_4709 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 0 & ... + g53391_4628 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 0 & ... + g53417_4431 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 0 & ... + g53392_4624 & 0 & ... & 0 & ... & 2 & [email protected] & 0 & ... & 0 & ... + g53414_4435 & 0 & ... & 0 & ... & 3 & [email protected] & 0 & ... & 0 & ... lrrrrr li i : & & & & & + 5000/2.5/1.0 & 0.10 & 0.13 & 0.01 & 0.00 & 0.01 + c i : & & & & & + 4250/1.0/1.4 & 0.10 & -0.28 & 0.09 & -0.02 & -0.03 + 5000/2.5/1.0 & 0.07 & -0.20 & 0.09 & -0.01 & 0.00 + 5500/4.0/0.6 & 0.07 & -0.11 & 0.07 & 0.00 & 0.00 + o i : & & & & & + 4250/1.0/1.4 & 0.08 & -0.12 & 0.09 & -0.01 & -0.03 + 5000/2.5/1.0 & 0.06 & -0.06 & 0.08 & -0.01 & -0.02 + na i : & & & & & + 4250/1.0/1.4 & 0.17 & 0.09 & -0.01 & -0.07 & 0.01 + 5000/2.5/1.0 & 0.11 & 0.07 & -0.01 & -0.03 & 0.01 + 5500/4.0/0.6 & 0.10 & 0.06 & -0.01 & 0.00 & 0.00
+ mg i : & & & & & + 4250/1.0/1.4 & 0.12 & 0.01 & 0.00 & -0.04 & 0.00 + 5000/2.5/1.0 & 0.10 & 0.05 & -0.01 & -0.03 & 0.00 + 5500/4.0/0.6 & 0.16 & 0.06 & -0.03 & -0.02 & 0.00 + al i : & & & & & + 4250/1.0/1.4 & 0.09 & 0.07 & 0.00 & -0.03 & 0.00 + 5000/2.5/1.0 & 0.06 & 0.06 & 0.00 & -0.01 & 0.01 + si i : & & & & & + 4250/1.0/1.4 & 0.09 & -0.12 & 0.06 & -0.03 & -0.04 + 5000/2.5/1.0 & 0.10 & -0.07 & 0.05 & -0.03 & -0.04 + 5500/4.0/0.6 & 0.08 & 0.00 & 0.01 & -0.02 & -0.01 + k i : & & & & & + 4250/1.0/1.4 & 0.26 & 0.14 & -0.01 & -0.16 & 0.00 + 5000/2.5/1.0 & 0.26 & 0.12 & -0.06 & -0.07 & 0.00 + ca i : & & & & & + 4250/1.0/1.4 & 0.18 & 0.12 & -0.02 & -0.10 & 0.00 + 5000/2.5/1.0 & 0.13 & 0.09 & -0.03 & -0.05 & 0.00 + 5500/4.0/0.6 & 0.16 & 0.08 & -0.05 & -0.02 & -0.01 + sc ii : & & & & & + 4250/1.0/1.4 & 0.20 & -0.02 & 0.07 & -0.12 & -0.06 + 5000/2.5/1.0 & 0.12 & -0.01 & 0.08 & -0.05 & -0.06 + 5500/4.0/0.6 & 0.10 & 0.00 & 0.08 & -0.01 & -0.05 + ti i : & & & & & + 4250/1.0/1.4 & 0.16 & 0.15 & 0.00 & -0.10 & 0.01 + 5000/2.5/1.0 & 0.08 & 0.12 & 0.00 & -0.03 & 0.01 + 5500/4.0/0.6 & 0.08 & 0.09 & 0.00 & -0.01 & 0.01 + v i : & & & & & + 4250/1.0/1.4 & 0.22 & 0.18 & 0.01 & -0.12 & 0.00 + 5000/2.5/1.0 & 0.09 & 0.15 & 0.00 & -0.02 & 0.01 + cr i : & & & & & + 4250/1.0/1.4 & 0.14 & 0.10 & 0.00 & -0.09 & 0.01 + 5000/2.5/1.0 & 0.08 & 0.08 & 0.00 & -0.03 & 0.01 + mn i : & & & & & + 4250/1.0/1.4 & 0.20 & 0.08 & 0.01 & -0.07 & 0.00 + 5000/2.5/1.0 & 0.12 & 0.10 & 0.00 & -0.03 & 0.01 + fe i : & & & & & + 4250/1.0/1.4 & 0.15 & 0.03 & 0.02 & -0.09 & -0.02 + 5000/2.5/1.0 & 0.11 & 0.08 & 0.00 & -0.06 & 0.00 + 5500/4.0/0.6 & 0.14 & 0.08 & -0.01 & -0.03 & 0.00 + fe ii : & & & & & + 4250/1.0/1.4 & 0.09 & -0.20 & 0.13 & -0.04 & -0.09 + 5000/2.5/1.0 & 0.09 & -0.14 & 0.11 & -0.04 & -0.08 + 5500/4.0/0.6 & 0.08 & -0.05 & 0.09 & -0.03 & -0.05 + co i : & & & & & + 4250/1.0/1.4 & 0.09 & 0.04 & 0.04 & -0.02 & -0.04 + 5000/2.5/1.0 & 0.05 & 0.10 & 0.01 & 0.00 & -0.01 + ni i : & & & & & + 4250/1.0/1.4 & 0.20 & 0.01 & 0.04 & -0.10 & -0.04 + 5000/2.5/1.0 & 0.11 & 0.07 & 0.01 & -0.06 & 0.00 + 5500/4.0/0.6 & 0.13 & 0.05 & 0.00 & -0.03 & -0.01 + cu i : & & & & & + 4250/1.0/1.4 & 0.26 & 0.05 & 0.05 & -0.10 & -0.03 + 5000/2.5/1.0 & 0.13 & 0.11 & 0.01 & -0.03 & 0.00 + zn i : & & & & & + 4250/1.0/1.4 & 0.06 & -0.03 & 0.05 & -0.01 & -0.02 + y ii : & & & & & + 4250/1.0/1.4 & 0.07 & -0.03 & 0.08 & -0.03 & -0.06 + zr i : & & & & & + 4250/1.0/1.4 & 0.13 & 0.20 & 0.01 & -0.09 & 0.00 + 5000/2.5/1.0 & 0.05 & 0.15 & 0.00 & -0.01 & 0.01 + ba ii : & & & & & + 4250/1.0/1.4 & 0.23 & 0.01 & 0.06 & -0.19 & -0.08 + 5000/2.5/1.0 & 0.20 & 0.02 & 0.04 & -0.16 & -0.07 + 5500/4.0/0.6 & 0.21 & 0.03 & 0.03 & -0.10 & -0.07 + la ii : & & & & & + 4250/1.0/1.4 & 0.08 & 0.01 & 0.08 & -0.04 & -0.07 + 5000/2.5/1.0 & 0.05 & 0.01 & 0.08 & -0.01 & -0.07 + eu ii : & & & & & + 4250/1.0/1.4 & 0.09 & -0.02 & 0.08 & -0.06 & -0.07 + 5000/2.5/1.0 & 0.06 & 0.00 & 0.08 & -0.02 & -0.06 + lccccc li & 3 & + 1.10 & 0.17 & 0.13 & ... + c & 6 & + 1.30 & 0.49 & 0.13 & 1.00 @xmath10 0.37 + o & 25 & + 0.19 & 0.18 & 0.12 & 0.48 @xmath10 0.10 + na & 25 & + 0.21 & 0.13 & 0.10 & 0.23 @xmath10 0.11 + mg & 24 & + 0.36 & 0.09 & 0.12 & 0.18 @xmath10 0.07 + al & 11 & + 0.24 & 0.10 & 0.10 & 0.11 @xmath10 0.10 + si & 24 & + 0.28 & 0.14 & 0.17 & 0.19 @xmath10 0.12 + k & 11 & 0.17 & 0.30 & 0.29 & 0.52 @xmath10 0.25 + ca & 25 & + 0.43 & 0.05 & 0.11 & 0.03 @xmath10 0.04 + sc & 25 & + 0.05 & 0.16 & 0.15 & 0.37 @xmath10 0.10 + ti & 25 & + 0.20 & 0.08 & 0.11 & 0.08 @xmath10 0.07 + v & 21 & + 0.11 & 0.14 & 0.17 & 0.09 @xmath10 0.08 + cr & 23 & 0.13 & 0.09 & 0.09 & 0.05 @xmath10 0.05 + mn & 13 & 0.27 & 0.11 & 0.11 & 0.03 @xmath10 0.08 + co & 17 & + 0.04 & 0.05 & 0.10 & 0.06 @xmath10 0.04 + ni & 25 & + 0.01 & 0.06 & 0.09 & 0.16 @xmath10 0.03 + cu & 21 & + 0.07 & 0.14 & 0.21 & 0.19 @xmath10 0.10 + zn & 8 & + 0.46 & 0.16 & 0.22 & 0.29 @xmath10 0.10 + y & 3 & 0.22 & 0.04 & 0.15 & 0.06 @xmath10 0.08 + zr & 10 & 0.14 & 0.25 & 0.16 & 0.61 @xmath10 0.24 + ba & 25 & + 0.34 & 0.12 & 0.19 & 0.03 @xmath10 0.09 + la & 14 & + 0.20 & 0.10 & 0.15 & 0.02 @xmath10 0.10 + eu & 11 & + 0.31 & 0.15 & 0.15 & 0.31 @xmath10 0.11 +
|
we present abundance ratios for 23 elements with respect to fe in a sample of stars with a wide range in luminosity from luminous giants to stars near the turnoff in a globular cluster .
our sample of 25 stars in m71 includes 10 giant stars more luminous than the red horizontal branch ( rhb ) , 3 hb stars , 9 giant stars less luminous than the rhb , and 3 stars near the turnoff .
the analyzed spectra , obtained with hires at the keck observatory , are of high dispersion ( r=@xmath0=35,000 ) .
we find that the neutron capture , the iron peak and the @xmath1element abundance ratios show no trend with @xmath2 , and low scatter around the mean between the top of the rgb and near the main sequence turnoff .
the @xmath1elements mg , ca , si and ti are overabundant relative to fe . the anti - correlation between o and na abundances , observed in other metal poor globular clusters , is detected in our sample and extends to the main sequence . a statistically significant correlation between al and na abundances
is observed among the m71 stars in our sample , extending to @xmath3 , fainter than the luminosity of the rgb bump in m5 .
lithium is varying , as expected , and zr may be varying from star to star as well .
m71 appears to have abundance ratios very similar to m5 whose bright giants were studied by @xcite , but seems to have a smaller amplitude of star - to - star variations at a given luminosity , as might be expected from its higher metallicity .
both extremely o poor , na rich stars and extremely o rich , na poor stars such as are observed in m5 and in m13 are not present in our sample of m71 stars .
the results of our abundance analysis of 25 stars in m71 provide sufficient evidence of abundance variations at unexpectedly low luminosities to rule out the mixing scenario . either alone or ,
even more powerfully , combined with other recent studies of c and n abundances in m71 stars , the existence of such abundance variations can not be reproduced within the context of our current understanding of stellar evolution .
|
the investigations of the properties of hadrons under extreme conditions have been in the focus of much attention for many years .
such investigations are very important in the study of the internal structure of the dense astrophysical objects like neutron stars .
the formation of neutron stars is influenced by all four known fundamental interactions .
hence , understanding of their nature can help us in the course of unification of all fundamental forces within a common theoretical framework , which is one of the biggest challenges for physics .
the recent observation of massive neutron stars with roughly twice the solar mass @xcite has stimulated the focuses on the equation of state of the dense nuclear matter ( see for instance @xcite ) .
however , the expected appearance of hyperons at about two times nuclear density , called `` hyperon puzzle '' remains an unresolved mystery in neutron stars ( concerning the appearance of hyperons in neutron stars see for example @xcite ) .
it has also found that @xmath0 isobars appear at a density of the order of @xmath5 times nuclear matter saturation density , and a `` @xmath0 puzzle '' exists , similar to the hyperon puzzle if the potential of the @xmath0 in nuclear matter is close to the one indicated by the experimental data @xcite .
more theoretical and experimental investigations on the properties of strange and non - strange light baryons in dense medium are needed to solve such puzzles . from the experimental side ,
the bound nuclear systems with one , two or three units of strangeness are poorly known compared to that of the non - strange states like nucleons .
the large production probability of various hyperon - antihyperon pairs in antiproton collisions will provide opportunities for series of new studies on the behavior of the systems containing two or even more units of strangeness at the @xmath4anda experiment at fair . by the progresses
made , it will be possible to study the in - medium properties of the doubly strange @xmath6-hypernuclei as well as the multi - strange @xmath7 , @xmath8 and @xmath3 systems in near future @xcite .
from the theoretical side , the effects of nuclear medium on the physical parameters of the nucleon have been widely investigated in the literature ( see for instance @xcite and references therein ) .
but , we have only a few studies dedicated to the in - medium properties of hyperons and decuplet baryons in the literature ( for instance see @xcite ) . in the present study
, we investigate the impact of nuclear matter on some spectroscopic parameters of the @xmath0 , @xmath1 , @xmath2 and @xmath3 decuplet baryons .
in particular , we calculate the mass and residue as well as the scalar and vector self energies of these baryons using the well established in - medium qcd sum rule approach .
we compare the in - medium results with those obtained at @xmath9 or vacuum and find the corresponding shifts . to remove the contributions of the negative energy particles , we work at the rest frame of the nuclear matter and fix the three momentum of the particles under consideration .
in this section we aim to construct sum rules for the mass , residue and vector self energy of the decuplet baryons and numerically analyze the obtained results . to this end and in accordance with the general philosophy of the qcd sum rule approach ,
we start with a correlation function as the building block of the method : @xmath10|\psi_0\rangle},\ ] ] where @xmath11 is the four momentum of the decuplet ( @xmath12 ) baryon , @xmath13 is the ground state of the nuclear matter and @xmath14 is the interpolating current of the @xmath12 baryon .
the general form of the interpolating current for decuplet baryons in a compact form reads , @xmath15 where @xmath16 are color indices , @xmath17 is the charge conjugation operator and @xmath18 is the normalization constant .
the quark content and value of @xmath18 for different members are given in table i @xcite .
.the value of the normalization constant @xmath18 and the quark flavors @xmath19 , @xmath20 , @xmath21 for the decuplet baryons . [ cols="^,^,^,^,^",options="header " , ] from this table , first of all , we see that our predictions on the masses in vacuum are in good consistencies with the average experimental data presented in pdg @xcite .
the masses obtained in the nuclear medium show negative shifts for all decuplet baryons . from the values of the scalar self energy ( @xmath22 ) , demonstrating the shifts in the masses due to finite density , we deduce that the maximum shift in the masses , due to the nuclear medium , with amount of @xmath23 belongs to the @xmath0 baryon and its minimum , @xmath24 , corresponds to the @xmath3 state .
this is an expected result since the @xmath0 state have the same quark content as the nuclear medium and is more affected by the nuclear matter . when going from @xmath0 to @xmath3 the up and down quarks
are replaced with the strange quark .
the @xmath3 state , having three @xmath25 quarks , is less affected by the medium .
the small shifts in the parameters of @xmath3 may be attributed to the intrinsic strangeness in the nucleons . in the case of the residues ,
our predictions in vacuum are overall comparable with those obtained in @xcite within the errors .
the small differences may be linked to different input parameters used in these works .
the values of residues are also considerably affected by the medium .
the shift in the residue of @xmath0 with amount of @xmath26 is maximum .
the residue of @xmath3 again is minimally affected by the medium with amount of roughly @xmath27 .
the value of vector self energy is considerably large in all decuplet channels .
it is again systematically reduced when going from the @xmath0 to @xmath3 baryon .
our results may be confronted with the experimental data of @xmath4anda collaboration at fair and nica facility .
however , we should remark that those experiments correspond to heavy ion collisions and not exactly to a nuclear medium .
hence , the appropriate way to make such comparison would be to present sum rules at finite density but where the density is introduced through the baryonic chemical potential .
this offers the possibility of exploring a wide range of densities .
we worked with the nuclear matter density since the the in - medium condensates are available as functions of nuclear matter density not chemical potential and we extracted the zero - density ( vacuum ) sum rules , as a means of normalizing the finite density sum rules , to compare the results with the available experimental data and other theoretical predictions in vacuum .
k. a. thanks dou university for the financial support through the grant bap 2015 - 16-d1-b04 .
n. u. bastian and d. blaschke , `` towards a new quark - nuclear matter eos for applications in astrophysics and heavy - ion collisions , '' j. phys .
ser . * 668 * , no . 1 , 012042 ( 2016 ) [ arxiv:1511.05881 [ nucl - th ] ] . e. s. fraga , a. kurkela and a. vuorinen , `` interacting quark matter equation of state for compact stars , '' astrophys . j. * 781 * , no .
2 , l25 ( 2014 ) [ arxiv:1311.5154 [ nucl - th ] ] . k. kim , h. k. lee and j. lee , `` compact star matter : eos with new scaling law , '' arxiv:1607.03235 [ nucl - th ] . m. drews and w. weise , `` from asymmetric nuclear matter to neutron stars : a functional renormalization group study , '' phys .
c * 91 * ( 2015 ) no.3 , 035802 doi:10.1103/physrevc.91.035802 [ arxiv:1412.7655 [ nucl - th ] ] .
s. weissenborn , d. chatterjee and j. schaffner - bielich , `` hyperons and massive neutron stars : vector repulsion and su(3 ) symmetry , '' phys .
c * 85 * , no .
6 , 065802 ( 2012 ) erratum : [ phys .
c * 90 * , no . 1 , 019904 ( 2014 ) ] [ arxiv:1112.0234 [ astro-ph.he ] ] . a. drago , a. lavagno , g. pagliara and d. pigato , `` early appearance of isobars in neutron stars , '' phys .
c * 90 * , no .
6 , 065809 ( 2014 ) doi:10.1103/physrevc.90.065809 [ arxiv:1407.2843 [ astro-ph.sr ] ] .
b. singh _ et al .
_ [ panda collaboration ] , `` study of doubly strange systems using stored antiprotons , '' nucl .
a * 954 * , 323 ( 2016 ) .
t. d. cohen , r. j. furnstahl , d. k. griegel and x. m. jin , `` qcd sum rules and applications to nuclear physics , '' prog .
phys . * 35 * , 221 ( 1995 ) [ hep - ph/9503315 ] .
e. g. drukarev and e. m. levin , `` the qcd sum rules and nuclear matter , '' jetp lett .
* 48 * , 338 ( 1988 ) [ pisma zh .
fiz . * 48 * , 307 ( 1988 ) ] .
k. azizi and n. er , `` properties of nucleon in nuclear matter : once more , '' eur . phys .
j. c * 74 * , 2904 ( 2014 ) [ arxiv:1401.1680 [ hep - ph ] ] .
k. azizi , n. er and h. sundu , `` positive and negative parity hyperons in nuclear medium , '' phys .
d * 92 * , no . 5 , 054026 ( 2015 ) [ arxiv:1506.02183 [ hep - ph ] ] .
x. m. jin and r. j. furnstahl , `` qcd sum rules for lambda hyperons in nuclear matter , '' phys .
c * 49 * , 1190 ( 1994 ) .
y. chung , h. g. dosch , m. kremer and d. schall , `` baryon sum rules and chiral symmetry breaking , '' nucl .
b * 197 * , 55 ( 1982 ) .
r. thomas , t. hilger and b. kampfer , `` four - quark condensates in nucleon qcd sum rules , '' nucl .
a * 795 * , 19 ( 2007 ) [ arxiv:0704.3004 [ hep - ph ] ] .
x. m. jin , t. d. cohen , r. j. furnstahl and d. k. griegel , `` qcd sum rules for nucleons in nuclear matter .
2 . , '' phys . rev .
c * 47 * , 2882 ( 1993 ) .
z. g. wang , `` analysis of the doubly heavy baryons in the nuclear matter with the qcd sum rules , '' eur .
j. c * 72 * , 2099 ( 2012 ) [ arxiv:1205.0605 [ hep - ph ] ] .
v. m. belyaev and b. l. ioffe , `` determination of the baryon mass and baryon resonances from the quantum - chromodynamics sum rule .
strange baryons , '' sov . phys .
jetp * 57 * , 716 ( 1983 ) [ zh .
* 84 * , 1236 ( 1983 ) ] ; b. l. ioffe , `` qcd at low energies , '' prog .
* 56 * , 232 ( 2006 ) [ hep - ph/0502148 ] .
j. m. alarcon , j. martin camalich and j. a. oller , `` the chiral representation of the @xmath28 scattering amplitude and the pion - nucleon sigma term , '' phys .
d * 85 * , 051503 ( 2012 ) [ arxiv:1110.3797 [ hep - ph ] ] .
a. w. thomas , p. e. shanahan and r. d. young , `` strangeness in the nucleon : what have we learned ?
, '' nuovo cim .
c * 035n04 * , 3 ( 2012 ) [ arxiv:1202.6407 [ nucl - th ] ] .
s. dinter , v. drach and k. jansen , `` dark matter search and the scalar quark contents of the nucleon , '' int .
j. mod .
e * 20 * , 110 ( 2011 ) [ arxiv:1111.5426 [ hep - lat ] ] . j. m. alarcon , l. s. geng , j. martin camalich and j. a. oller , `` the strangeness content of the nucleon from effective field theory and phenomenology , '' phys .
b * 730 * , 342 ( 2014 ) [ arxiv:1209.2870 [ hep - ph ] ] .
x. m. jin , m. nielsen , t. d. cohen , r. j. furnstahl and d. k. griegel , `` qcd sum rules for nucleons in nuclear matter .
3 . , '' phys . rev .
c * 49 * , 464 ( 1994 ) .
f. x. lee , `` predicative ability of qcd sum rules for decuplet baryons , '' phys .
c * 57 * , 322 ( 1998 ) [ hep - ph/9707332 ] .
k. azizi and g. bozkir , `` decuplet baryons in a hot medium , '' eur .
j. c * 76 * , no .
10 , 521 ( 2016 ) [ arxiv:1606.05452 [ hep - ph ] ] .
|
the decuplet baryons , @xmath0 , @xmath1 , @xmath2 and @xmath3 , are studied in nuclear matter by using the in - medium qcd sum rules . by fixing the three momentum of the particles under consideration at the rest frame of the medium ,
the negative energy contributions are removed .
it is obtained that the parameters of the @xmath0 baryon are more affected by the medium against the @xmath3 state , containing three strange quarks , whose mass and residue do not affected by the medium , considerably .
we also find the vector and scalar self energies of these baryons in nuclear matter . by the recent progresses at @xmath4anda experiment at fair and nica facility it may be possible to study the in - medium properties of such states even the multi - strange @xmath2 and @xmath3 systems in near future .
|
a one - time pad can famously be used only once @xcite , i.e. , a secret key as long as the message is needed to encrypt it with information / theoretic security .
but this does not hold anymore if the honest players can use quantum technologies to communicate .
a quantum key distribution ( qkd ) protocol @xcite allows players to expand an initial short secret key , and thus encrypt messages that are longer than the length of the original key . instead of first expanding a key , and then using it for encryption , one can also swap the order if the initial key is long enough : one first encrypts a message , then recycles the key .
this is possible due to the same physical principles as qkd : quantum states can not be cloned , so if the receiver holds the exact cipher that was sent , the adversary can not have a copy , and thus does not have any information about the key either , so it may be reused .
this requires the receiver to verify the authenticity of the message received , and if this process fails , a net key loss occurs the same happens in qkd : if an adversary tampers with the communication , the players have to abort and also lose some of the initial secret key .
some ideas for recycling encryption keys using quantum ciphers were already proposed in 1982 @xcite .
many years later , gottesman @xcite and then damgrd et al .
@xcite showed how to encrypt a classical message in a quantum state and recycle the key . at roughly the same time , the first protocol for authenticating quantum messages was proposed by barnum et al .
@xcite , who also proved that quantum authentication necessarily encrypts the message as well .
their protocol was then adapted to recycle key by oppenheim and horodecki @xcite .
but the security definitions in these initial works on quantum authentication have a major flaw : they do not consider the possibility that an adversary may hold a purification of the quantum message that is encrypted .
this was corrected by hayden , leung and mayers @xcite , who give a composable security definition for quantum authentication with key recycling .
they then show that the family of protocols from @xcite are secure , and prove that one can recycle part of the key if the message is accepted .
the security proof from @xcite does however not consider all possible environments .
starting in works by simmons in the 80 s and then stinson in the 90 s ( see , for example , @xcite ) the classical literature on authentication studies two types of attacks : _ substitution attacks _ where the adversary obtains a valid pair of message and cipher and attempts to substitute the cipher with one that will decode to a different message and _ impersonation attacks _ where the adversary directly sends a forged cipher to the receiver , without knowledge of a valid message / cipher pair . to the best of our knowledge
, there is no proof showing that security against impersonation attacks follows from security against substitution attacks , hence the literature analyzes both attacks separately .
this is particularly important in the case of composable security , which aims to prove the security of the protocol when used in any arbitrary environment , therefore also in an environment that first sends a forged cipher to the receiver , learns wether it is accepted or rejected , then provides a message to the sender to be authenticated , and finally obtains the cipher for this message .
this is all the more crucial when key recycling is involved , since the receiver will already recycle ( part of ) the key upon receiving the forged cipher , which is immediately given to the environment .
the work of hayden et al .
@xcite only considers environments that perform substitution attacks
i.e. , first provide the sender with a message , then change the cipher , and finally learn the outcome of the authentication as well as receive the recycled key .
hence they do not provide a complete composable security proof of quantum authentication .
more recently , alternative security definitions for quantum authentication have been proposed , both without @xcite and with @xcite key recycling .
these still only consider substitution attacks , and furthermore , they are , strictly speaking , not composable . while it is possible to prove that these definitions imply security in a composable framework ( if one restricts the environment to substitution attacks ) , the precise way in which the error @xmath0 carries over to the framework has not been worked out in any of these papers .
if two protocols with composable errors @xmath0 and @xmath1 are run jointly ( e.g. , one is a subroutine of the other ) , the error of the composed protocol is bounded by the sum of the individual errors , @xmath2 . if a security definition does not provide a bound on the composable error , then one can not evaluate the new error after composition .. ] for example , quantum authentication with key recycling requires a backwards classical authentic channel , so that the receiver may tell the sender that the message was accepted , and allow her to recycle the key .
the error of the complete protocol is thus the sum of errors of the quantum authentication and classical authentication protocols .
definitions such as those of @xcite are not sufficient to directly obtain a bound on the error of such a composed protocol . in the other direction
, it is immediate that if a protocol is @xmath0/secure according to the composable definition used in this work , then it is secure according to @xcite with the same error @xmath0 .
more precisely , proving that the quantum authentication scheme constructs a secure channel is sufficient to satisfy @xcite
i.e. , the ideal functionality is a secure channel which only allows the adversary to decide if the message is delivered , but does not leak any information about the message to the adversary except its length ( confidentiality ) , nor does it allow the adversary to modify the message ( authenticity ) . and proving that the scheme constructs a secure channel that additionally generates fresh secret key is sufficient to satisfy the definition of _ total authentication _ from @xcite .
garg et al .
@xcite also propose a definition of _ total authentication with key leakage _ , which can be captured in a composable framework by a secure channel that generates fresh key and leaks some of it to the adversary .
this is however a somewhat unnatural ideal functionality , since it requires a deterministic leakage function , which may be unknown or not exist , e.g. , the bits leaked can depend on the adversary s behavior this is the case for the _ trap code _
@xcite , discussed in .
the next natural step for players in such a situation is to extract a secret key from the partially leaked key , and thus the more natural ideal functionality is what one obtains after this privacy amplification step @xcite : a secure channel that generates fresh secret key , but where the key generated may be shorter than the key consumed .
the ideal functionality used in the current work provides this flexibility : the amount of fresh key generated is a parameter which may be chosen so as to produce less key than consumed , the same amount , or even more .
hence , with one security definition , we encompass all these different cases no key recycling , partial key recycling , total key recycling , and even a net gain of secret key . furthermore , having all these notions captured by ideal functionalities makes for a particularly simple comparison between the quite technical definitions appearing in @xcite . in this work we use the abstract cryptography ( ac )
framework @xcite to model the composable security of quantum authentication with key recycling .
ac views cryptography as a resource theory : a protocol constructs a ( strong ) resource given some ( weak ) resources . for example , the quantum authentication protocols that we analyze construct a secure quantum channel a channel that provides both _
confidentiality _ and _ authenticity _ that additionally produces fresh key . in order to construct this resource ,
we require shared secret key , an insecure ( noiseless ) quantum channel and a backwards authentic classical channel .
these are all resources , that may in turn be constructed from weaker resources , e.g. , the classical authentic channel can be constructed from a shared secret key and an insecure channel , and noiseless channels are constructed from noisy channels . due to this constructive aspect of the framework , it is also called _ constructive cryptography _ in the literature @xcite . although this approach is quite different from the universal composability ( uc ) framework @xcite , in the setting considered in this work
with one dishonest player and where recipients are denoted by classical strings the two frameworks are essentially equivalent and the same results could have been derived with a quantum version of uc @xcite . in uc , the constructed resource would be called _ ideal functionality _ , and the resources used in the construction are setup assumptions .
we thus first formally define the ideal resource constructed by the quantum authentication protocol with key recycling the secure channel that produces fresh key mentioned in this introduction as well as the resources required by this construction .
we then prove that a family of quantum authentication protocols proposed by barnum et al .
@xcite satisfy this construction , i.e. , no distinguisher ( called environment in uc ) can distinguish the real system from the ideal resource and simulator except with an advantage @xmath0 that is exponentially small in the security parameter .
this proof considers all distinguishers allowed by quantum mechanics , including those that perform impersonation attacks .
we show that in the case where the message is accepted , every bit of key may be recycled . and
if the message is rejected , one may recycle all the key except the bits used to one - time pad the cipher .
we prove that this is optimal , i.e. , an adversary may obtain all non / recycled bits of key .
this improves on previous results , which recycled less key .
more specifically , hayden et al .
@xcite , while also analyzing protocols from @xcite , only recycle part of the key in case of an accept , and lose all the key in case of a reject .
garg et al .
@xcite propose a new protocol , which they prove can recycle all of the key in the case of an accept , but do not consider key recycling in the case of a reject either .
the protocols we analyze are also more key efficient than that of @xcite .
we give two instances which need @xmath3 bits of initial secret key , instead of the @xmath4 required by @xcite , where @xmath5 is the length of the message and @xmath0 is the error .
note that the family of protocols for which we provide a security proof is a subset of the ( larger ) family introduced in @xcite .
more precisely , barnum et al .
@xcite define quantum authentication protocols by composing a quantum one - time pad and what they call a _ purity testing code _ which , with high probability , will detect any noise that may modify the encoded message whereas we require a stricter notion , a _ strong purity testing code _ which , with high probability , will detect any noise .
this restriction on the family of protocols is necessary to recycle all the key .
in fact , there exists a quantum authentication scheme , the _ trap code _
@xcite , which is a member of the larger class from @xcite but not the stricter class analyzed here , and which leaks part of the key to the adversary , even upon a successful authentication of the message this example is discussed in .
we then give two explicit instantiations of this family of quantum authentication protocols .
the first is the construction used in @xcite , which requires an initial key of length @xmath6 , where @xmath5 is the length of the message and @xmath7 is the security parameter , and has error @xmath8 .
the second is an explicit unitary @xmath9/design @xcite discovered by chau @xcite , which requires @xmath10 bits of initial key bits of key , but we show that some of the unitaries are redundant when used for quantum authentication and can be dropped . ] and has error @xmath11 .
both constructions have a net loss of @xmath12 bits of key if the message fails authentication .
since several other explicit quantum authentication protocols proposed in the literature are instances of this family of schemes , our security proof is a proof for these protocols as well this is discussed further in .
finally , we show how to construct the resources used by the protocol from nothing but insecure noisy channels and shared secret key , and calculate the joint error of the composed protocols .
we also show how to compensate for the bits of key lost in the construction of the backwards authentic channel , so that the composed protocol still has a zero net key consumption if no adversary jumbles the communication . in
we give a brief introduction to the main concepts of ac , which are necessary to understand the notion of cryptographic construction and corresponding security defintion . a more extended introduction to ac
is provided in . in we
then define the resources constructed and used by a quantum authentication scheme with key recycling .
we introduce the family of protocols from @xcite that we analyze in this work , and then prove that they construct the corresponding ideal secure channel .
we also prove that the number of recycled bits is optimal . in we
show how to construct the various resources used by the quantum authentication protocol , and put the pieces together to get a security statement for the joint protocol that constructs the secure quantum channel that generates fresh key from nothing but noisy insecure channels and shared secret key .
finally , in we discuss the relation between some quantum authentication schemes that have appeared in the literature and those analyzed here , as well as some open problems .
as already mentioned in , the ac framework @xcite models cryptography as a resource theory . in this section we give a brief overview of how these constructive statements are formalized .
we illustrate this with an example taken from @xcite , namely authentication of classical messages with message authentication codes ( mac ) . an expanded version of this introduction to ac is provided in .
[ .5][c ] 1.4 ( ch ) at ( 0,0 ) ; at ( ch.north west ) insecure channel @xmath13 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( alice.center ) to ( -,0 ) to ( -,- ) ; ( , - ) to ( , 0 ) to ( bob.center ) ; [ .5][c ] 1.4 ( ch ) at ( 0,0 ) ; at ( ch.north west ) authentic channel @xmath14 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( alice.center ) to ( 0,0 ) to node ( handle ) ( 330:2 * ) ; ( 2*,0 ) to ( bob.center ) ; ( -,0 ) to ( -,- ) ; ( handle.center |- 0,- ) to node[auto , swap , pos=.3 ] @xmath15 ( handle.center ) ; [ .5][c ] .2 ( keybox ) at ( 0,0 ) ; ( key ) at ( 0,- ) key ; at ( keybox.north west ) secret key @xmath16 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( up ) at ( 0 , ) ; ( down ) at ( 0,- ) ; ( alice |- up ) to node[auto ] req .
( keybox.west |- up ) ; ( bob |- up ) to node[auto , swap ] req .
( keybox.east |- up ) ; ( key.west |- down ) to node[auto , pos=.8 ] @xmath17 ( alice |- down ) ; ( key.east |- down ) to node[auto , swap , pos=.8 ] @xmath17 ( bob |- down ) ; [ .5][c ] 1.5.2 ( ch ) at ( 0,0 ) ; at ( ch.north west ) insecure channel @xmath13 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( alice.center ) to ( -,0 ) to ( -,- ) ; ( , - ) to ( , 0 ) to ( bob.center ) ; ( keybox ) at ( 0 , ) ; ( key ) at ( 0,u- ) key ; at ( keybox.north west ) secret key @xmath16 ; ( keylabel ) at ( keybox.north ) ; ( up ) at ( 0,+ ) ; ( down ) at ( 0,u- ) ; ( alice |- up ) to node[auto , pos=.4 ] req .
( keybox.west |- up ) ; ( bob |- up ) to node[auto , swap , pos=.4 ] req .
( keybox.east |- up ) ; ( key.west |- down ) to node[auto , pos=.8 ] @xmath17 ( alice |- down ) ; ( key.east |- down ) to node[auto , swap , pos=.8 ] @xmath17 ( bob |- down ) ; ( comp ) ; at ( comp.north west ) composed resource @xmath18 ; in an @xmath7 player setting , a _ resource _ is an object with @xmath7 interfaces , that allows every player to input messages and receive other messages at her interface .
the objects depicted in are examples of resources .
the insecure channel in allows alice to input a message at her interface on the left and allows bob to receive a message at his interface on the right .
eve can intercept alice s message and insert a message of her choosing at her interface .
the authentic channel resource depicted in also allows alice to send a message and bob to receive a message , but eve s interface is more limited than for the insecure channel : she can only decide if bob receives the message or not , but not tamper with the message being sent . the key resource drawn in provides each player with a secret key when requested . if two resources @xmath16 and @xmath13 are both available to the players , we write @xmath18 for the resource resulting from their parallel composition
this is to be understood as the resources being merged into one : the interfaces belonging to player @xmath19 are simultaneously accessible to her as one new interface , which we depict in .
_ converters _ capture operations that a player might perform locally at her interface .
for example , if the players share a key resource and an insecure channel , alice might decide to append a mac to her message .
this is modeled as a converter @xmath20 that obtains the message @xmath21 at the outside interface , obtains a key at the inside interface from a key resource @xmath16 and sends @xmath22 on the insecure channel @xmath13 , where @xmath23 is taken from a family of strongly @xmath9/universal hash functions @xcite .
we illustrate this in .
converters are always drawn with rounded corners . if a converter @xmath24 is connected to the @xmath19 interface of a resource @xmath25 , we write @xmath26 or @xmath27 for the new resource obtained by connecting the two . and @xmath28 interfaces on the left and converters at the @xmath29 interface on the right , though there is no mathematical difference between @xmath26 and @xmath27 . ]
2.89 .75 i.846 ( a1 ) at ( - , ) ; ( a2 ) at ( -,0 ) ; ( a3 ) at ( -,- ) ; ( a ) at ( -,0 ) @xmath30 + @xmath31 ; at ( a.north west ) @xmath32 ; ( alice ) at ( -,0 ) ; ( b1 ) at ( , ) ; ( b2 ) at ( , 0 ) ; ( b3 ) at ( , - ) ; ( b ) at ( , 0 ) @xmath33 + @xmath34 ; at ( b.north west ) @xmath35 ; ( bob ) at ( , 0 ) ; ( keybox ) at ( 0 , ) ; ( keyinnerboxleft ) at ( -i , ) ; ( keyinnerboxright ) at ( i , ) ; ( key ) at ( 0 , ) key ; at ( keybox.north west ) secret key @xmath16 ; ( channel ) at ( 0,- ) ; at ( channel.north west ) insecure channel @xmath13 ; ( eveleft ) at ( -.4,-1.75 ) ; ( everight ) at ( .4,-1.75 ) ; ( ajunc ) at ( eveleft |- a3 ) ; ( bjunc ) at ( everight |- b3 ) ; ( key ) to node[auto , swap , pos=.15,yshift=-2 ] @xmath17 ( a1 ) ; ( key ) to node[auto , pos=.15,yshift=-2 ] @xmath17 ( b1 ) ; ( a1 ) to node[auto , pos=.5 ] req .
( keyinnerboxleft ) ; ( b1 ) to node[auto , swap , pos=.5 ] req .
( keyinnerboxright ) ; ( alice.center ) to node[auto , pos=.4 ] @xmath21 ( a2 ) ; ( b2 ) to node[auto , pos=.75 ] @xmath36 ( bob.center ) ; ( a3 ) to ( ajunc.center ) to node[pos=.8,auto , swap ] @xmath37 ( eveleft.center ) ; ( everight.center ) to node[pos=.2,auto , swap ] @xmath38 ( bjunc.center ) to ( b3 ) ; a protocol is then defined by a set of converters , one for every honest player .
another type of converter that we need is a _ filter_. the resources illustrated in depict a setting with an adversary that has some control over these resources . for a cryptographic protocol to be useful it is not sufficient to provide guarantees on what happens when an adversary is present
, one also has to provide a guarantee on what happens when no adversary is present , e.g. , if no adversary tampers with the message on the insecure channel , then bob will receive the message that alice sent .
we model this setting by covering the adversarial interface with a filter that emulates an honest behavior . in
we draw an insecure and an authentic channel with filters @xmath39 and @xmath40 that transmit the message to bob . in the case of the insecure channel
, one may want to model an honest noisy channel when no adversary is present .
this is done by having the filter @xmath39 add some noise to the message .
a dishonest player removes this and has access to a noiseless channel as in .
[ .5][c ] 1.4 ( ch ) at ( 0,0 ) ; at ( ch.north west ) insecure channel @xmath13 ; ( fil ) at ( 0,- ) ; at ( fil.north west ) @xmath39 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( alice.center ) to ( -,0 ) to ( -,- ) ; ( -,- ) to ( , - ) ; ( , - ) to ( , 0 ) to ( bob.center ) ; [ .5][c ] 1.4 ( ch ) at ( 0,0 ) ; at ( ch.north west ) authentic channel @xmath14 ; ( fil ) at ( 0,- ) ; at ( fil.north west ) @xmath40 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( alice.center ) to ( bob.center ) ; ( -,0 |- ch.center ) to ( -,0 |- fil.center ) ; ( , 0 |- fil.center ) to node[auto , swap , pos=.45,xshift=-1 ] @xmath41 ( , 0 |- ch.center ) ; we use the term _ filtered resource _ to refer to a pair of a resource @xmath25 and a filter @xmath39 , and often write @xmath42
. such an object can be thought of as having two modes : it is characterized by the resource @xmath43 when no adversary is present and by the resource @xmath25 when the adversary is present .
the final object that is required by the ac framework to define the notion of construction and prove that it is composable , is a ( pseudo/)metric defined on the space of resources that measures how close two resources are . in the following , we use a distinguisher based metric , i.e. , the maximum advantage a distinguisher has in guessing whether it is interacting with resource @xmath25 or @xmath44 , which we write @xmath45 .
more specifically , let @xmath46 be a distinguisher , and le @xmath47 $ ] and @xmath48 $ ] be the binary random variables corresponding to @xmath46 s output when connected to @xmath25 and @xmath44 , respectively . then the distinguishing advantage between @xmath25 and @xmath44
is defined as @xmath49 = 0 \right ] - \pr\left [ \ad[\as ] = 0 \right ] \right|\,.\ ] ] since we study information / theoretic security in this work , the supremum is taken over the set of all possible distinguishers allowed by quantum mechanics .
this is discussed further in .
we are now ready to define the security of a cryptographic protocol .
we do so in the three player setting , for honest alice and bob , and dishonest eve .
thus , in the following , all resources have three interfaces , denoted @xmath50 , @xmath28 and @xmath29 , and a protocol is then given by a pair of converters @xmath51 for the honest players .
we refer to @xcite for the general case , when arbitrary players can be dishonest .
[ def : security ] let @xmath52 be a protocol and @xmath53 and @xmath54 denote two filtered resources .
we say that _ @xmath55 constructs @xmath56 from @xmath57 within @xmath0 _ , which we write @xmath58 , if the two following conditions hold : 1 . [
eq : def.cor ] we have @xmath59 2 . [ eq : def.sec ] there exists a converter @xmath60 which we call simulator such that @xmath61 if it is clear from the context what filtered resources @xmath57 and @xmath62 are meant , we simply say that @xmath55 is @xmath0/secure .
the first of these two conditions measures how close the constructed resource is to the ideal resource in the case where no malicious player is intervening , which is often called _ correctness _ in the literature .
the second condition captures _ security _ in the presence of an adversary .
for example , to prove that the mac protocol constructs an authentic channel from an insecure channel and a secret key within @xmath0 , we need to find a converter @xmath63 such that the real system @xmath64 can not be distinguished from the ideal system @xmath65 with advantage greater than @xmath0 . the simulator @xmath63 drawn in satisfies this requirement if the family of hash functions @xmath66 is @xmath0/almost strongly @xmath9/universal @xcite .
-1.1 .75 ( channel ) at ( 0 , ) ; at ( channel.north west ) authentic channel @xmath14 ; ( alice ) at ( - , ) ; ( bob ) at ( , ) ; ( sim ) at ( 0 , ) ; at ( sim.north west ) @xmath67 ; ( sleft ) at ( -.809 , ) ; ( sright ) at ( .809 , ) ; ( sim.north ) to ( sim.south ) ; ( ajunc ) at ( sleft |- alice ) ; ( bjunc ) at ( sright |- bob ) ; ( alice.center ) to node[pos=.15,auto ] @xmath21 ( .4 , ) to node[pos=.54 ] ( ejunc ) + ( 160:-.8 ) ; ( ajunc.center ) to node[pos=.63,auto , swap ] @xmath21 ( sleft ) ; ( 1.2 , ) to node[pos=.75,auto ] @xmath68 ( bob.center ) ; ( sright ) to node[pos=.4,auto , swap ] @xmath15 ( ejunc.center ) ; ( sltext ) at ( -.809 , ) @xmath69 ; ( srtext ) at ( .809 , ) @xmath70 + @xmath71 ; ( eveleft ) at ( sleft |- 0 , ) ; ( everight ) at ( sright |- 0 , ) ; ( sleft ) to node[pos=.75,auto , swap ] @xmath37 ( eveleft.center ) ; ( everight.center ) to node[pos=.25,auto , swap ] @xmath38 ( sright ) ; ( key ) at ( .15,+.8 ) key ; ( key ) to ( sleft ) ; [ rem : efficiency ] the protocols and simulators discussed in this work are all efficient .
the protocols we consider are either trivially efficient or taken from other work , in which case we refer to these other works for proofs of efficiency .
the efficiency of the simulator used to prove the security of quantum authentication has been analyzed in @xcite .
all other simulators used in the security proofs run the corresponding honest protocols , and are thus efficient because the protocols are .
we therefore do not discuss efficiency any further in this work .
we start with some technical preliminaries in , where we introduce ( strong ) purity testing codes , which are a key component of the family of quantum authentication protocols of @xcite . in
we give a constructive view of quantum authentication with key recycling : we define the resource that such a protocol is expected to construct , as well as the resources that are required to achieve this . in
we describe the family of protocols that we analyze in this work , along with a variant in which the order of the encryption and encoding operations has been swapped , which we prove to be equivalent . in
we give a security proof for the family of quantum authentication protocols defined earlier . and
in we show that the number of recycled key bits is optimal .
finally , in we give two explicit constructions of purity testing codes and get the exact parameters of the quantum authentication protocols with these codes . [ [ pauli - operators . ] ] pauli operators .
+ + + + + + + + + + + + + + + + to denote a pauli operator on @xmath7 qubits we write either @xmath72 or @xmath73 , where @xmath21 and @xmath74 are @xmath7-bit strings indicating in which positions bit and phase flips occur , and @xmath75 is the concatenation of @xmath21 and @xmath74 , which is used when we do not need to distinguish between @xmath21 and @xmath74 .
two pauli operators @xmath76 and @xmath77 with @xmath78 and @xmath79 commute ( anti / commute ) if the symplectic inner product @xmath80 is @xmath41 ( is @xmath81 ) , where @xmath82 is the scalar product of the vectors and the arithmetic is done modulo @xmath9 . hence , for any @xmath83 and @xmath73 @xmath84
we use several times the following equality @xmath85 where @xmath86 means that all bits of the string @xmath87 are @xmath41 .
[ [ purity - testing - code . ] ] purity testing code .
+ + + + + + + + + + + + + + + + + + + + an error correcting code ( ecc ) that encodes an @xmath5 qubit message in a @xmath88 qubit code word is generally defined by an isomorphism from @xmath89 to @xmath90 . in this work
we define an ecc by a unitary @xmath91 .
the code word for a state @xmath92 is obtained by appending a @xmath7 qubit state @xmath93 to the message , and applying @xmath94 , i.e. , the encoding of @xmath92 is @xmath95 .
we do not need to use the decoding properties of eccs in this work , we only use the them to detect errors , i.e. , given a state @xmath96 , we apply the inverse unitary @xmath97 and measure the last @xmath7 qubits to see if they are @xmath93 or not .
the first property we require of our codes , is that they map any pauli error @xmath73 into another pauli error @xmath98 , i.e. , @xmath99 for some global phase @xmath100 .
this is always the case for any @xmath94 that can be implemented with clifford operators . in particular , all stabilizer codes have this property , which are used in @xcite to define purity testing codes .
note that the mapping from @xmath87 to @xmath101 is a permutation on the set of indices @xmath102 that depends only on the choice of code .
a code will detect an error @xmath73 if @xmath103 for @xmath104 , where @xmath72 acts on the first @xmath5 qubits and @xmath105 on the last @xmath7 .
measuring these last qubits would yield the syndrome @xmath106 , since @xmath105 flips the bits in the positions corresponding to the bits of @xmath106 . and
an error @xmath73 will act trivially on the message if @xmath107 . in particular ,
if @xmath108 , then this error will not be detected , but not change the message either . for a code indexed by a key @xmath17 , we denote by @xmath109 the set of pauli errors that are not detected by this code , and by @xmath110 we denote the undetected errors which act trivially on the message .
a purity testing code is a set of codes @xmath111 such that when a code @xmath112 is selected uniformly at random , it will detect with high probability all pauli errors which act non / trivially on the message .
a purity testing code with error @xmath0 is a set of codes @xmath111 , such that for all pauli operators @xmath73 , @xmath113 as mentioned in , we use a stricter definition of purity testing code in this work .
we require that all non / identity paulis get detected with high probability , even those that act trivially on the message .
intuitively , the reason for this is that , with the original definition of purity testing , if the adversary introduces some noise @xmath73 , by learning whether the message was accepted or not , she will learn whether that error acts trivially on the message or not , and thus learn something about the ecc used .
this means that the adversary learns something about the key used to choose the ecc , and hence it can not be recycled in its entirety .
a strong purity testing code with error @xmath0 is a set of codes @xmath111 , such that for all non / identity pauli operators @xmath73 , @xmath114 in we provide explicit constructions of strong purity testing codes .
the main result in this paper is a proof that the family of quantum authentication protocols of barnum et al .
@xcite restricted to strong purity testing codes can be used to construct a secure quantum channel that additionally produces fresh key , which we denote @xmath115 and illustrate in .
such a filtered resource can be thought of as a merger between a secure channel
i.e. , a resource that transmits a message from alice to bob , only allowing eve to learn that a message has been transmitted ( or the length of the message , if this is not predetermined ) and also allowing her to prevent bob from getting the message and a secret key resource
i.e. , a resource as depicted in , that provides both honest players with a new secret key , but where eve has additional power to control the length of the secret key distributed .
the resource @xmath116 is parametrized by @xmath117 values . the length of the message @xmath5 that it can transmit , and two values @xmath118 denoting the length of the secret key that is generated .
if eve is not active , both players get a fresh secret key of length @xmath119 .
however , eve can prevent the players from recycling all the key , so a key of length @xmath120 is generated instead .
furthermore , eve can prevent alice from learning whether the cipher was received by bob , so that she can not recycle any key
if alice were to recycle her key before bob receives the cipher , eve could use the information from the recycled key to modify the cipher without being detected .
a more detailed description of this filtered resource is provided in the caption of .
.6 ( channel ) at ( 0,0 ) ; at ( channel.north west ) secure channel @xmath116 ; ( aliceup ) at ( - , ) ; ( alicemiddle ) at ( -,0 ) ; ( alicedown ) at ( -,- ) ; ( bobup ) at ( , ) ; ( bobmiddle ) at ( , 0 ) ; ( bobdown ) at ( , - ) ; ( eveleft ) at ( -,- ) ; ( evecenter ) at ( 0,- ) ; ( everight ) at ( , - ) ; ( alicedown ) to node[auto , pos=.1,yshift=-1 ] @xmath121 ( -,- ) to + ( 340:2 * ) ; ( + , - ) to node[auto , pos=.8,yshift=-1 ] @xmath122 ( bobdown ) ; ( eveleft |- alicedown ) to ( eveleft ) ; ( key ) at ( , 0 ) key ; ( key ) to node[auto , pos=.2 ] @xmath17 ( bobmiddle ) ; ( key ) to ( , 0 ) to node[auto , swap , pos=.1 ] @xmath17 node[pos=.53 ] ( handle ) + ( 200:2 * ) ; ( -,0 ) to ( alicemiddle ) ; ( everight ) to node[pos=.19,xshift=-1,auto , swap ] @xmath15 node[circle , fill , pos=.65,inner sep=1.5 ] ( key ) ; ( evecenter ) to node[pos=.17,xshift=-1,auto , swap ] @xmath15 ( handle.center ) ; ( aliceup ) to node[auto , yshift=-1 ] req .
( channel.west |- aliceup ) ; ( bobup ) to node[auto , swap , yshift=-1 ] req .
( channel.east |- bobup ) ; ( channels ) at ( , 0 ) ; at ( channels.north west ) secure channel @xmath116 ; ( aliceups ) at ( -+ , ) ; ( alicemiddles ) at ( -+,0 ) ; ( alicedowns ) at ( -+,- ) ; ( bobups ) at ( + , ) ; ( bobmiddles ) at ( + , 0 ) ; ( bobdowns ) at ( + , - ) ; ( fil ) at ( , - ) ; at ( fil.east ) @xmath123 ; ( evelefts ) at ( -+,- ) ; ( evecenters ) at ( , - ) ; ( everights ) at ( + , - ) ; ( alicedowns ) to node[auto , pos=.07,yshift=-1 ] @xmath121 node[auto , pos=.93,yshift=-1 ] @xmath121 ( bobdowns ) ; ( evelefts |- alicedowns ) to ( evelefts |- fil.north ) ; ( keys ) at ( + , 0 ) key ; ( keys ) to node[auto , pos=.2 ] @xmath17 ( bobmiddles ) ; ( keys ) to node[auto , swap , pos=.1 ] @xmath17 ( alicemiddles ) ; ( everights |- fil.north ) to node[pos=.22,xshift=-1,auto , swap ] @xmath41 ( keys ) ; ( fil.north ) to node[pos=.18,xshift=-1,auto , swap ] @xmath41 ( , 0 ) ; at ( everights |- bobdowns ) ; ( aliceups ) to node[auto , yshift=-1 ] req .
( channels.west |- aliceups ) ; ( bobups ) to node[auto , swap , yshift=-1 ] req .
( channels.east |- bobups ) ; to construct @xmath115 , the quantum authentication protocol will use a shared secret key to encrypt and authenticate the message .
this means that the players must share a secret key resource @xmath124 ( ) , is empty , this resource has a trivial empty filter , which we do not write down . ]
where @xmath125 denotes the length of the key this resource provides .
they also need to share an insecure quantum channel , which is used to send the message , and is illustrated in without a filter and in with a filter .
the authentication protocol we consider is designed to catch any error , so if it is used over a noisy channel , it will always abort .
we thus assume that the players share a noiseless channel , which we denote @xmath126 , i.e. , @xmath13 is controlled by the adversary as in .
but if no adversary is present , the filter @xmath127 is noiseless . finally , the players need a backwards authentic channel , that can send one bit of information from bob to alice .
this is required so that alice may learn whether the message was accepted and recycle the corresponding amount of key . the authentic channel and its filter @xmath128 are drawn in figures [ fig : channel.dishonest.authentic ] and [ fig : channel.honest.authentic ] .
putting this together in the case of an active adversary , we get , where the converters for alice s and bob s parts of the quantum authentication protocol are labeled @xmath129 and @xmath130 , respectively . 1.5 ( auth ) at ( 0,0 ) ; at ( auth.north west ) authentic channel @xmath14 ; ( ch ) at ( -,- ) ; at ( ch.north ) insecure ch . @xmath13 ; ( key ) at ( 0 , ) ; at ( key.north west ) secret key @xmath124 ; ( prota ) at ( -,0 ) ; at ( prota.north west ) @xmath131 ; ( a1 ) at ( - , ) ; ( a2 ) at ( -,0 ) ; ( a3 ) at ( -,- ) ; ( protb ) at ( , 0 ) ; at ( protb.north east ) @xmath132 ; ( b1 ) at ( , ) ; ( b2 ) at ( , 0 ) ; ( b3 ) at ( , - ) ; ( aliceup ) at ( - , ) ; ( alicemiddle ) at ( -,0 ) ; ( alicedown ) at ( -,- ) ; ( bobup ) at ( , ) ; ( bobmiddle ) at ( , 0 ) ; ( bobdown ) at ( , - ) ; ( eveleftl ) at ( --,- ) ; ( eveleftr ) at ( -+,- ) ; ( everightl ) at ( -,- ) ; ( everightr ) at ( + , - ) ; ( alicedown ) to node[auto , pos=.4 ] @xmath121 ( a3 ) ; ( a3 ) to ( eveleftl |- alicedown ) to ( eveleftl ) ; ( eveleftr ) to ( eveleftr |- bobdown ) to ( b3 ) ; ( b3 ) to node[auto , pos=.8 ] @xmath133 ( bobdown ) ; ( aliceup ) to node[auto , pos=.3 ] req . ( a1 ) ; ( bobup ) to node[auto , swap , pos=.3 ] req . ( b1 ) ; ( a2 ) to node[auto , swap , pos=.8 ] @xmath134 ( alicemiddle ) ; ( b2 ) to node[auto , pos=.6 ] @xmath135 ( bobmiddle ) ; ( key ) at ( 0 , ) key ; ( key ) to node[auto , swap , pos=.2 ] @xmath17 ( a1 ) ; ( key ) to node[auto , pos=.2 ] @xmath17 ( b1 ) ; ( kleft ) at ( -1.1 , ) ; ( kright ) at ( 1.1 , ) ; ( a1 ) to node[auto , pos=.5 ] req .
( kleft ) ; ( b1 ) to node[auto , swap , pos=.5 ] req .
( kright ) ; ( b2 ) to ( , 0 ) to node[pos=.53 ] ( handle ) + ( 200:2 * ) ; ( -2*,0 ) to ( a2 ) ; ( everightr |- bobmiddle ) to ( everightr ) ; ( everightl ) to node[auto , swap , pos=.1,xshift=-2 ] @xmath15 ( handle.center ) ; a protocol @xmath136 is then a quantum authentication protocol with error @xmath137 if it constructs @xmath138 from @xmath139 within @xmath137 , i.e. , @xmath140 in we describe the protocol , and in we prove that is satisfied and provide the parameters @xmath141 .
the family of quantum authentication protocols from @xcite consists in first encrypting the message to be sent with a quantum one - time pad , then encoding it with a purity testing code and a random syndrome .
we do the same , but with a strong purity testing code .
we also extend the protocol so that the players recycle all the key if the message is accepted , and the key used to select the strong purity testing code if the message is rejected .
so that alice may also recycle the key , bob uses the backwards authentic classical channel to notify her of the outcome .
we refer to this as the `` encrypt / then / encode '' protocol , the details of which are provided in . quantum authentication encrypt / then / encode 1 .
alice and bob obtain uniform keys @xmath17 , @xmath87 , and @xmath106 from the key resource , where @xmath17 is long enough to choose an element from a strong purity testing code that encodes @xmath5 qubits in @xmath88 qubits , @xmath87 is @xmath142 bits and @xmath106 is @xmath7 bits . 2 .
alice encrypts the message @xmath143 she receives with a quantum one - time pad using the key @xmath87 .
she then appends an @xmath7 qubit state @xmath144 , and encodes the whole thing with a strong purity testing code , obtaining the cipher @xmath145 .
3 . alice sends @xmath146 to bob on the insecure channel .
bob receives a message @xmath147 , he applies @xmath148 , decrypts the @xmath50 part and measures the @xmath149 part in the computational basis .
if the result of the measurement is @xmath106 , he accepts the message and recycles @xmath17 , @xmath87 and @xmath106 .
if the result is not @xmath106 , he rejects the message , and recycles @xmath17 .
bob sends alice a bit on the backwards authentic channel to tell her if he accepted or rejected the message .
when alice receives bob s bit , she either recycles all the keys or only @xmath17 .
alternatively , one may perform the encoding and encryption in the opposite order : alice first encodes her message with the strong purity testing code with syndrome @xmath41 , then does a quantum one - time pad on the resulting @xmath88 qubit state .
this `` encode / then / encrypt '' protocol is described in . quantum authentication encode / then / encrypt 1 .
alice and bob obtain uniform keys @xmath17 and @xmath87 from the key resource , where @xmath17 is long enough to choose an element from a strong purity testing code that encodes @xmath5 qubits in @xmath88 qubits and @xmath87 is @xmath6 bits long . 2 .
alice appends a @xmath7 qubit state @xmath150 to the message @xmath143 she receives , encodes it with a strong purity testing code chosen according to the key @xmath17 , and encrypts the whole thing with a quantum one - time pad using the key @xmath87 .
she thus obtains the cipher @xmath151 .
3 . alice sends @xmath146 to bob on the insecure channel .
4 . bob receives a message @xmath147 , he applies @xmath73 , then @xmath148 , and measures the @xmath149 part in the computational basis .
if the result of the measurement is @xmath41 , he accepts the message and recycles @xmath17 and @xmath87 .
otherwise , he rejects the message , and recycles @xmath17 .
bob sends alice a bit on the backwards authentic channel to tell her if he accepted or rejected the message .
when alice receives bob s bit , she either recycles all the keys or only @xmath17 .
the encode / then / encrypt protocol uses @xmath7 bits more key , and since these bits are not recycled in case of a reject , it is preferable to use the encrypt / then / encode protocol .
these protocols are however identical : no external observer can detect which of two is being run .
this holds , because the encode / then / encrypt protocol performs phase flips on a syndrome that is known to be in a computational basis state @xmath152 .
thus , they have no effect and can be skipped .
likewise , bob performs phase flips on @xmath149 before measuring in the computational basis he might as well skip these phase flips , since they have no effect either .
we formalize this statement by proving ( in ) that the converters corresponding to the two different protocols are indistinguishable .
this result is similar in spirit to proofs that some prepare / and / measure quantum key distribution ( qkd ) protocols are indistinguishable from entanglement - based qkd protocols , and thus security proofs for one are security proofs for the other @xcite .
since these two protocols are indistinguishable , we provide a security proof in for the encode / then / encrypt protocol .
however , in , when we count the number of bits of key consumed , we count those of the encrypt / then / encode protocol . [
lem : ete ] let @xmath153 and @xmath154 denote the pairs of converters modeling alice s and bob s behavior in the encrypt / then / encode and encode / then / encrypt protocols , respectively .
then @xmath155 we start with alice s part of the protocol .
let @xmath156 and @xmath131 receive keys @xmath17 , @xmath87 and @xmath106 as in the protocol from , as well as an extra key @xmath74 of length @xmath7 that is needed by @xmath131 , since it requires more key .
the distinguisher prepares a state @xmath157 , and sends the @xmath50 part to the system .
@xmath156 outputs @xmath158 where in the last line we used .
this is exactly the state output by @xmath131 if when receiving the key @xmath159 , the protocol uses the pauli @xmath98 for the quantum one - time pad . for bob s part of the protocol ,
let the distinguisher prepare a state @xmath160 and send the @xmath161 part to the system .
the subnormalized state held jointly by @xmath162 and the distinguisher after decoding and performing the measurement is given by @xmath163 we again obtain the state that is jointly held by @xmath132 and the distinguisher if when receiving the key @xmath159 , the protocol uses the pauli @xmath98 for the quantum one - time pad .
[ rem : classicalmessage ] if part of the message is classical
i.e. , it is diagonal in the computational basis and known not to have a purification held be the distinguisher then running the same proof as , one can show that it is sufficient to perform bit flips on that part of the message , the phase flips are unnecessary .
this is the case with the protocol from , that generates a key @xmath21 locally and sends it to bob using a quantum authentication scheme .
we use this to save some bits of key in the bounds from and .
suppose that there exists a strong purity testing code @xmath111 of size @xmath164 and with error @xmath0 that encodes an @xmath5 qubit message in an @xmath88 qubit cipher . and let @xmath165 denote alice and bob s converters when running the encode / then / encrypt protocol from .
more precisely , if @xmath129 receives a message at its outer interface , it requests a key from the key resource , encrypts the message as described and sends the cipher on the insecure channel .
it may receive three symbols from the backwards authentic channel : an error @xmath166 , in which case it does not recycle any key , a message @xmath41 saying that @xmath130 did not receive the correct state , in which case it recycles the part of the key of length @xmath167 used to choose the code , or a message @xmath81 saying that @xmath130 did receive the correct state , in which case it recycles all the key .
if @xmath129 first receives a message on the backward authentic channel before receiving a message to send , it will not recycle any key .
similarly , when @xmath130 receives a cipher on the insecure channel , it requests a key from the key resource , performs the decryption , outputs either the message or an error depending on the result of the decryption , and sends this result back to @xmath129 on the authentic channel .
the real system consisting of these converters and the resources they use is depicted in .
[ thm : auth ] let @xmath168 denote the protocol described above
. then @xmath168 constructs a secure channel that recycles key , @xmath169 , given an insecure quantum channel @xmath126 , a backward authentic channel @xmath170 and a secret key @xmath171 , i.e. , @xmath172 with @xmath173 , where @xmath0 is the error of the strong purity testing code . in order to prove this theorem
, we need to find a simulator such that the real and ideal systems are indistinguishable except with advantage @xmath174 .
the simulator that we use is illustrated in .
.6 ( channel ) at ( 0,0 ) ; at ( channel.north west ) secure channel @xmath116 ; ( sim ) at ( 0,- ) ; at ( sim.north east ) @xmath175 ; ( aliceup ) at ( - , ) ; ( alicemiddle ) at ( -,0 ) ; ( alicedown ) at ( -,- ) ; ( bobup ) at ( , ) ; ( bobmiddle ) at ( , 0 ) ; ( bobdown ) at ( , - ) ; ( eveleftl ) at ( -3*,- ) ; ( eveleftr ) at ( -,- ) ; ( everightl ) at ( , - ) ; ( everightr ) at ( 3*,- ) ; at ( -2.5*,- ) @xmath176 ; ( alicedown ) to node[auto , pos=.1,yshift=-1 ] @xmath121 ( -,- ) to + ( 340:2 * ) ; ( + , - ) to node[auto , pos=.8,yshift=-1 ] @xmath122 ( bobdown ) ; ( eveleftl |- alicedown ) to ( eveleftl |-sim.north ) ; ( eveleftl ( key ) at ( , 0 ) key ; ( key ) to node[auto , pos=.2 ] @xmath17 ( bobmiddle ) ; ( key ) to ( -+,0 ) to node[auto , swap , pos=.1 ] @xmath17 node[pos=.53 ] ( handle ) + ( 200:2 * ) ; ( --,0 ) to ( alicemiddle ) ; ( everightl |- sim.north ) to node[pos=.2,xshift=-1,auto , swap ] @xmath15 node[circle , fill , pos=.61,inner sep=1.5 ] ( key ) ; ( eveleftr ) to node[pos=.08,xshift=-1,auto , swap ] @xmath15 ( handle.center ) ; ( everightr |- sim.south ) to node[auto , pos=.6 ] @xmath15 ( everightr ) ; ( everightl ) to node[auto , pos=.5,xshift=-1,swap ] @xmath177 ( everightl |- sim.south ) ; ( aliceup ) to node[auto , yshift=-1 ] req .
( channel.west |- aliceup ) ; ( bobup ) to node[auto , swap , yshift=-1 ] req .
( channel.east |- bobup ) ; it is trivial to show that correctness holds with error @xmath41 , namely that @xmath178 we now prove the case of security , i.e. , @xmath179 the real and ideal systems , drawn in figures [ fig : auth.real ] and [ fig : auth.ideal ] have @xmath180 inputs .
the distinguisher thus has the choice between @xmath181 possible orders for providing inputs .
however , most of these orders are redundant and do not need to be analyzed .
providing the requests for the secret keys before they are ready is pointless .
so it is sufficient to look at the case where these requests are made as soon as the keys are available for recycling , i.e. , after bob has received the message from alice and after alice has received the confirmation from bob .
what is more , neither sending alice an error on the backwards authentic channel nor allowing her to get bob s confirmation will help either way , since the distinguisher already knows what output alice will produce , so we can completely ignore this input . that leaves only @xmath9 in - ports , and thus @xmath9 orders to analyze : 1 .
the distinguisher first inputs a message at alice s interface , gets the cipher at eve s interface , inputs a possibly modified cipher at eve s interface , gets the output at bob s interface , and requests the recycled key .
2 . the distinguisher first inputs a fake cipher at eve s interface ,
gets the output at bob s interface , makes a request for his recycled key , then inputs a message at alice s interface and receives the cipher for that message .
we start with the first case , the initial message is sent to alice .
the distinguisher prepares a message @xmath182 and inputs the @xmath183 part at alice s interface .
the ideal channel then notifies the simulator that a message has been input .
the simulator prepares a maximally entangled state @xmath176 of dimension @xmath184 and outputs the @xmath185 register at eve s interface .
the distinguisher now holds a bipartite state in @xmath186 , to which it applies a unitary @xmath187 . without loss of generality ,
one may write the unitary as @xmath188 , where @xmath189 are paulis acting on the cipher register @xmath185 and @xmath190 act on the distinguisher s internal memory @xmath29 . the resulting state in the @xmath185 register is input back in the @xmath29 interface .
the simulator now measures @xmath191 in the bell basis defined by the projectors @xmath192 .
if the outcome is @xmath193 where @xmath194 it tells the channel that the cipher was not modified . in which case the contents of the register @xmath183 is output at bob s interface with an @xmath195 flag . furthermore , it generates a fresh uniform key @xmath196 , where @xmath197 and @xmath198 .
if the outcome is @xmath199 , then the simulator notifies the channel to delete the message and output a @xmath200 flag , as well as prepare only the shorter key @xmath17 .
the distinguisher then sends a request to obtain the fresh key .
so the final state held by the distinguisher interacting with the ideal system is @xmath201
\\ { } + \sum_{j \neq 0 } \proj{{\textsf{rej } } } \tensor \tau^{k } \tensor e^e_j \rho^e \hconj{\left(e^e_j\right)}\,,\end{gathered}\ ] ] where @xmath202 and @xmath203 are fully mixed states and @xmath204{\proj{\psi}^{me}}$ ] .
one could append states @xmath205 and @xmath206 in the @xmath200 branch of so that both terms have the same number of registers ; we omit them for simplicity . in the real system , for the secret key @xmath196
, the state before bob s measurement of the syndrome is given by @xmath207 where @xmath208 denotes the symplectic product defined in .
let @xmath209 be the set of indices @xmath210 such that the error @xmath211 produces a syndrome @xmath106 when code @xmath17 is used , i.e. , @xmath212 ( see )
. for @xmath213 , let @xmath214 then @xmath215 the next step in bob s protocol consists in measuring the syndrome .
if @xmath108 is obtained , he outputs the message as well as the key @xmath196 and a flag @xmath195 .
otherwise he deletes the message , outputs @xmath17 with the flag @xmath200
. the final state held be the distinguisher in this case is @xmath216 where we have used @xmath217 for some unitary @xmath218 . setting @xmath219 the distance between real and ideal systems may be written as @xmath220 @xmath221 and @xmath222 are both pure states , so from we bound their distance as @xmath223 where @xmath224 is the vector @xmath9/norm and we used the fact that @xmath225 . from jensen s inequality and using
we obtain @xmath226 finally , because the code is a strong purity testing code with error @xmath0 and that @xmath227 with @xmath228 , we get @xmath229 in the reject branch of the real system we have @xmath230 where we used again .
thus @xmath231 putting all this together we get @xmath232 we now consider the second case : the distinguisher first prepares a state @xmath233 and inputs the @xmath185 part at eve s interface , then obtains the output at bob s interface
. note that in the ideal case the channel always outputs a @xmath200 message at bob s interface .
thus , if the cipher is accepted by bob who outputs a state @xmath221 the distinguisher must be interacting with the real system and can already output this guess . in the case of a rejection , it now holds a bipartite system @xmath234 the recycled key @xmath235 and its purifying system @xmath29 .
it then applies an isometry @xmath236 to this system and inputs the @xmath183 part of the resulting state at alice s interface .
after which it obtains a cipher at eve s interface and holds the tripartite system @xmath237 the recycled key @xmath235 , the cipher @xmath185 and its internal memory @xmath29 .
we denote this state @xmath238 in the ideal case and @xmath239 in the real case , and we need to bound @xmath240 in a first step , we assume that the state @xmath233 prepared by the distinguisher is an antisymmetric fully entangled state , which we denote @xmath241 , where @xmath242 is the hamming weight of @xmath243 and @xmath244 is the string @xmath21 with all bits flipped . in the ideal case
the simulator notifies the channel to reject the cipher , and the state @xmath245 is output at bob s interface .
the distinguisher then holds @xmath246 . in the real case
, bob applies the decoding algorithm , i.e. , first a pauli @xmath247 , then a unitary @xmath248 and finally measures @xmath7 bits of the syndrome in the computational basis .
since the antisymmetric state is invariant under @xmath249 , one could equivalently apply the inverse operation , @xmath250 , to the @xmath29 system , i.e. , the state after bob s measurement is given by @xmath251 if @xmath108 bob accepts the cipher as being valid , which happens with probability @xmath252 , i.e. , @xmath253 .
in the case where @xmath254 , he deletes the cipher , so the remaining state is given by @xmath255 where @xmath256 @xmath235 is made public and the @xmath257 system is the part of the key kept secret by the players .
let @xmath258 denote the completely positive , trace / preserving ( cptp ) map consisting of the distinguisher s next step the isometry @xmath236 and the final operation of the ideal system deleting the message system @xmath183 that is input at alice s interface and outputting a fully mixed state @xmath259 .
let @xmath260 denote the cptp map consisting of the distinguisher s next step and the final operation of the real system encoding the message system @xmath183 according to the protocol and outputting the resulting cipher .
we have @xmath261 and @xmath262 .
thus , @xmath263 since @xmath264 . finally , note that we have @xmath265 for @xmath266{u \left
( \tau^k \tensor \tau^e\right ) \hconj{u}}$ ] , since the random pauli @xmath73 applied by the encryption algorithm completely decouples the cipher from @xmath234 .
putting this together , we get @xmath267 since a strong purity testing code will always have an error @xmath268 . the final case that remains to consider is when the distinguisher prepares a state @xmath233 that is not the antisymmetric state .
we will reduce this case to that of the entangled antisymmetric by using the entangled state @xmath269 to teleport the @xmath177 part of any state @xmath270 .
let the teleportation scheme be given by the projectors @xmath271 on @xmath272 which incur a pauli correction @xmath273 on the @xmath185 system , i.e. , @xmath274{\sum_a p^c_a \tensor m^{ec'}_a \left ( \proj{\psi^-}^{ce } \tensor
\proj{\psi}^{c'e'}\right ) p^c_a \tensor m^{ec'}_a } = \proj{\psi}_{ce ' } \,.\ ] ] so the distinguisher prepares an entangled state @xmath275 and the state it wishes to send to bob , @xmath270 .
it teleports the @xmath177 register to the @xmath185 register , and sends this to bob , who performs his decryption operation @xmath276 .
this results in the shared state @xmath277 \\ & \qquad = \frac{1}{2^{\nu+2m+2n } } \sum_{k,\ell , a } \proj{k,\ell \xor a } \tensor \big [ \left(\hconj{\left(u_k^c\right ) } p^c_\ell \tensor m^{ec'}_a\right)\\ & \qquad \qquad \qquad \qquad \left ( \proj{\psi^-}^{ce } \tensor \proj{\psi}^{c'e'}\right ) \left(p^c_\ell u^c_k \tensor m^{ec'}_a\right ) \big ] \\ & \qquad = \frac{1}{2^{\nu+2m+2n } } \sum_{k,\ell , a } x^l_a \proj{k,\ell } x^l_a \tensor \big [ \left(i^c \tensor m^{ec'}_a p^e_\ell u^e_k \right)\\ & \qquad \qquad \qquad \qquad \left ( \proj{\psi^-}^{ce } \tensor \proj{\psi}^{c'e'}\right ) \left(i^c \tensor \hconj{\left(u_k^e\right ) } p^e_\ell m^{ec'}_a\right ) \big]\,,\end{aligned}\ ] ] where @xmath278 flips the bits of @xmath87 in the positions where @xmath279 .
the teleportation of @xmath280 is thus equivalent to a measurement of the distinguisher s system followed by a correction of the secret key @xmath87 .
this may however be performed after bob measures the syndrome and accepts or rejects the cipher he received .
the probability of accepting the cipher is thus unchanged , and plugging in the result from the case where the distinguisher sends bob half of the anti / symmetric entangled state , we find that the state of the real system in the rejection branch after bob s operations is @xmath281 where @xmath282 .
and in the ideal system the state after the ideal channel outputs a rejection is @xmath283 .
we thus obtain the same bound on the distance between real and ideal systems as in the previous case .
it follows from that is also a proof of security for the encrypt / then / encode protocol from , i.e. , @xmath284 with @xmath173 .
thus , in the case where the message is not accepted by bob , @xmath12 bits of key are lost .
we prove here that this is optimal : one can not recycle any extra bit of key .
[ lem : auth.optimal ] there exists an adversarial strategy to obtain all the secret bits that are not recycled in the encrypt / then / encode protocol .
the distinguisher prepares epr pairs @xmath285 and provides the @xmath183 part to alice .
it then receives the cipher and thus holds the state @xmath286 which it keeps .
it then sends a bogus cipher to bob , and obtains the key @xmath17 after bob recycles it .
it applies the decoding unitary @xmath287 , measures the @xmath149 register to get the secret key @xmath106 and measures the joint @xmath288 register in the bell basis to get the secret key @xmath87 .
the protocols we have given in use strong purity testing codes , and the parameters of the key used , key recycled and error depend on the parameters of these codes . in this section we give two constructions of purity testing codes .
the first requires less initial secret key , the second has a better error parameter .
both have the same net consumption of secret key bits .
the first construction is from barnum et al .
@xcite . they give an explicit strong purity testing code with @xmath289 and @xmath290 . and
we remove one of the keys ( and thus increase the error ) , so as to get simpler final expressions . ] plugging this in the parameters from with the encrypt / then / encode protocol , we get the following .
[ cor : auth.1 ] the encrypt / then / encode protocol with the purity testing code of @xcite requires an initial key of length @xmath6 .
it recycles all bits if the message is accepted , and @xmath7 bits if the message is rejected .
the error is @xmath292 the second construction we give is based on an explicit purity testing code by chau @xcite though he does not name it this way .
chau @xcite finds a set of unitaries @xmath293 in dimension @xmath294 such that , if @xmath17 is chosen uniformly at random , any non / identity pauli is mapped to every non / identity pauli with equal frequency , i.e. , @xmath295 with @xmath296 and @xmath297 , @xmath298 where @xmath299 is some global phase .
we prove in , , that this is a strong purity testing code with @xmath300 for @xmath301 .
it also has @xmath302 , hence @xmath303 .
note that when composed with paulis as in the encode / then / encrypt protocol , @xmath304 is a unitary @xmath9/design @xcite .
it follows that any ( approximate ) unitary @xmath305/design is a good quantum authentication scheme ( see for a formal proof ) .
[ cor : auth.2 ] the encrypt / then / encode protocol with the purity testing code of @xcite requires an initial key of length @xmath10 .
it recycles all bits if the message is accepted , and @xmath306 bits if the message is rejected .
the error is @xmath307 .
we proved in that the quantum authentication protocols from figures [ fig : protocol.1 ] and [ fig : protocol.2 ] construct a secure channel that recycles key @xmath138 from a shared secret key @xmath308 , a noiseless insecure quantum channel @xmath309 and a backwards authentic classical channel @xmath128 , namely , @xmath310 in this section we show how to construct the required resources from nothing but shared secret key and noisy channels , then put it all together to get the exact bounds of the composed protocols .
we discuss in how to obtain the authentic channel : it can be constructed from a shared secret key and an insecure channel using any ( classical ) mac / type authentication scheme @xcite .
channels are however usually not noiseless .
this is solved by using error correction : an error correction code constructs a noiseless channel given a noisy channel ( with known noise ) , and is presented in .
were we to put things together at this point , we would construct the desired secure quantum channel from nothing but shared secret key and insecure noisy channels .
but even in the case where no adversary is present , we would still consume some secret key , because we do not recycle the key from the backwards authentic channel .
so in we construct shared secret key given a secure channel we simply share secret key using the channel .
combining all these pieces together , we obtain our secure quantum channel without any net consumption of key in the case where the adversary does not tamper with the messages .
the security of the composed scheme follows immediately from the security of each component and the composition theorem of ac @xcite . finally , in we discuss a setting in which the backwards authentic channel is not needed , thus allowing a more efficient use of the quantum channel
since we do not need it to distribute key , and can thus use all of its capacity to send messages .
we used classical authentication as an example in : an authentication protocol @xmath311 constructs an authentic channel @xmath128 given a ( noiseless ) insecure channel @xmath126 and a secret key @xmath312 , @xmath313 the corresponding real and ideal systems were illustrated in figures [ fig : classicalauth.real ] and [ fig : classicalauth.ideal ] , respectively . appending the mac @xmath314 to the message @xmath21 is sufficient to construct the authentic channel with error @xmath0 if the family of functions @xmath66 is @xmath0/almost strongly @xmath9/universal @xcite . in our case , bob only needs to send a @xmath81-bit message to alice .
if the key @xmath17 has length @xmath315 , a @xmath316/almost strongly @xmath9/universal family of functions for @xmath81-bit messages is given by @xmath317 , where @xmath318 are the first @xmath319 bits of @xmath17 and @xmath320 are the last @xmath319 bits .
[ lem : classicalauth ] the authentication scheme described above satisfies with @xmath321
. follows from ( * ? ? ?
* lemma 9 ) . both the classical authentication protocol discussed in and [ sec : construction.auth ] , as well as the quantum authentication protocol analyzed in produce an error message as soon as there is any disturbance on the channel .
realistic channels are naturally noisy , so for such protocols to even make sense , one needs an extra layer of error correction that is designed to correct the specific noise on the channel . here , we formalize this as a constructive statement .
let @xmath322 encode a message with an error correcting code ( ecc ) given by the completely positive , trace / preserving ( cptp ) map @xmath258 , and @xmath323 decode the message with the cptp map @xmath324 , as illustrated in . and let @xmath325 be the error of the ecc for noise given by a cptp map @xmath260 , i.e. , @xmath326 then @xmath327 constructs a noiseless ( insecure ) channel @xmath328 from a noisy ( insecure ) channel @xmath329 , if the filter @xmath39 introduces the noise @xmath260 , i.e. , @xmath330 [ .5][c ] 1.3 ( ch ) at ( 0,0 ) ; at ( ch.north ) insecure ch .
@xmath331 ; ( fil ) at ( 0,- ) ; ( noise ) at ( 0,- ) @xmath260 ; at ( fil.north west ) @xmath39 ; ( prota ) at ( -,0 ) ; ( pia ) at ( -,0 ) @xmath258 ; at ( prota.north west ) @xmath322 ; ( protb ) at ( , 0 ) ; ( pib ) at ( , 0 ) @xmath324 ; at ( protb.north west ) @xmath323 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( alice.center ) to ( pia ) ; ( pia ) to ( -,0 ) to ( -,- ) ; ( -,- ) to ( noise ) ; ( noise ) to ( , - ) ; ( , - ) to ( , 0 ) to ( pib ) ; ( pib ) to ( bob.center ) ; [ .5][c ] 1.3 ( ch ) at ( 0,0 ) ; at ( ch.north west ) insecure channel @xmath13 ; ( fil ) at ( 0,- ) ; at ( fil.north west ) @xmath127 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( alice.center ) to ( -,0 ) to ( -,- ) ; ( -,- ) to ( , - ) ; ( , - ) to ( , 0 ) to ( bob.center ) ; [ .5][c ] 1.4 ( ch ) at ( 0,0 ) ; at ( ch.north ) insecure ch . @xmath331 ; ( prota ) at ( -,0 ) ; ( pia ) at ( -,0 ) @xmath258 ; at ( prota.north west ) @xmath322 ; ( protb ) at ( , 0 ) ; ( pib ) at ( , 0 ) @xmath324 ; at ( protb.north west ) @xmath323 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( evel ) at ( -,- ) ; ( ever ) at ( , - ) ; ( alice.center ) to ( pia ) ; ( pia ) to ( -,0 ) to ( evel ) ; ( ever ) to ( , 0 ) to ( pib ) ; ( pib ) to ( bob.center ) ; [ .5][c ] 1.3 ( ch ) at ( 0,0 ) ; at ( ch.north west ) insecure channel @xmath13 ; ( fil ) at ( 0,- ) ; at ( fil.north west ) @xmath332 ; ( alice ) at ( -,0 ) ; ( bob ) at ( , 0 ) ; ( evel ) at ( -,- ) ; ( ever ) at ( , - ) ; ( siml ) at ( -,- ) @xmath258 ; ( simr ) at ( , - ) @xmath324 ; ( alice.center ) to ( -,0 ) to ( siml ) ; ( siml ) to ( evel ) ; ( ever ) to ( simr ) ; ( simr ) to ( , 0 ) to ( bob.center ) ; note that the resources @xmath13 and @xmath331 are of different dimension , since an error correcting code will map a quantum state to a new one of larger dimension . in this work
we generally we do not use different notation for channels of different dimensions , since the dimension is usually clear from the context , and we juste write @xmath13 for an insecure channel .
we distinguish a noiseless from a noisy channel by its filter , @xmath127 and @xmath39 , respectively .
[ lem : noiseless ] if the filter @xmath39 introduces noise given by a cptp map @xmath260 , and the encoding and decoding maps @xmath258 and @xmath324 satisfy , then the protocol @xmath333 that uses this ecc satisfies .
we need to prove that @xmath334 and @xmath335 for some simulator @xmath332 to satisfy the two conditions from .
these systems are drawn in .
one can easily check from the figure that the first condition holds because the ecc was designed to achieve exactly this . @xmath336 and @xmath337 are each a pair of channels , one performing the encoding operation @xmath258 , the other the decoding operation @xmath324 , so the second condition holds with distance @xmath41 .
naturally , only makes sense if there exists a code than can correct the errors introduced by @xmath260 , i.e. , if there exist maps @xmath338 that satisfy . in this work ,
when we talk about a noisy channel resource , we always mean such a channel that has non / zero capacity .
it is trivial to share secret key using a secure channel : alice generates a uniform string and sends it on the secure channel to bob .
if the channel can transmit an @xmath339 qubit state and we use it to share a @xmath340 bit key , then @xmath5 qubits can still be used to transmit a message .
this protocol is depicted in .
.6 ( channel ) at ( 0,0 ) ; at ( channel.north west ) secure ch . @xmath341 ; ( pia ) at ( -,0 ) ; at ( pia.north west ) @xmath342 ; at ( pia.south ) @xmath343 ; ( pib ) at ( , 0 ) ; at ( pib.north west ) @xmath344 ; ( aliceup ) at ( - , ) ; ( alicemiddle ) at ( -,0 ) ; ( alicedown ) at ( -,- ) ; ( bobup ) at ( , ) ; ( bobmiddle ) at ( , 0 ) ; ( bobdown ) at ( , - ) ; ( eveleft ) at ( -,- ) ; ( evecenter ) at ( 0,- ) ; ( everight ) at ( , - ) ; ( pia.east |- alicedown ) to node[auto , pos=.1,yshift=-1 ] @xmath345 ( -,- ) to + ( 340:2 * ) ; ( + , - ) to ( pib.west |- bobdown ) ; at ( pib.west |- bobdown ) @xmath133 ; ( eveleft |- alicedown ) to ( eveleft ) ; ( alicedown ) to node[auto , yshift=-1 ] @xmath121 ( pia.west |- alicedown ) ; ( pib.east |- bobdown ) to node[auto , pos=.5,yshift=-1 ] @xmath122 ( bobdown ) ; ( key ) at ( , 0 ) key ; ( key ) to node[auto , pos=.2 ] @xmath17 ( pib ) ; ( key ) to ( , 0 ) to node[auto , swap , pos=.1 ] @xmath17 node[pos=.53 ] ( handle ) + ( 200:2 * ) ; ( -,0 ) to ( pia ) ; ( pia ) to node[auto , swap , pos=.75,yshift=-1 ] @xmath346 ( alicemiddle ) ; ( pib ) to node[auto , pos=.5,yshift=-1 ] @xmath347 ( bobmiddle ) ; ( everight ) to node[pos=.19,xshift=-1,auto , swap ] @xmath15 node[circle , fill , pos=.65,inner sep=1.5 ] ( key ) ; ( evecenter ) to node[pos=.17,xshift=-1,auto , swap ] @xmath15 ( handle.center ) ; ( pia.east |- aliceup ) to node[auto , yshift=-1 ] req .
( channel.west |- aliceup ) ; ( pib.west |- bobup ) to node[auto , swap , yshift=-1 ] req .
( channel.east |- bobup ) ; ( aliceup ) to node[auto , yshift=-1 ] req .
( pia.west |- aliceup ) ; ( bobup ) to node[auto , swap , yshift=-1 ] req .
( pib.east |- bobup ) ; [ lem : morekey ] let @xmath348 denote the protocol described above .
it constructs a @xmath349 secure channel from a @xmath350 secure channel with no error , i.e. , @xmath351 this lemma trivially holds with an empty simulator .
if we compose all the protocols described above , we obtain the system depicted in .
let @xmath352 denote the composition of alice s converters and @xmath353 denote the composition of bob s converters .
we then immediately get that the @xmath52 constructs a secure channel with key recycling , @xmath349 , from secret keys @xmath308 and @xmath312 , and two noisy channels @xmath354 and @xmath355 with error @xmath356 , where the arrows are used to distinguisher the forward quantum channel and the backward classical channel , i.e , @xmath357 1.5 1 ( keylong ) at ( 0,3 * ) ; at ( keylong.north west ) secret key @xmath124 ; ( keyshort ) at ( 0,2 * ) ; at ( keyshort.north west ) secret key @xmath358 ; ( chback ) at ( 0 , ) ; at ( chback.north west ) insecure channel @xmath359 ; ( chforward ) at ( -,0 ) ; at ( chforward.north west ) ins .
@xmath360 ; ( peccbacka ) at ( - , ) ; at ( peccbacka.north ) @xmath361 ; ( peccbackb ) at ( , ) ; at ( peccbackb.north ) @xmath362 ; ( peccforwarda ) at ( -,0 ) ; at ( peccforwarda.north ) @xmath363 ; ( peccforwardb ) at ( , 0 ) ; at ( peccforwardb.north ) @xmath364 ; ( pcautha ) at ( --,3*/2 ) ; at ( pcautha.north ) @xmath20 ; ( pcauthb ) at ( + , 3*/2 ) ; at ( pcauthb.north ) @xmath365 ; ( pqautha ) at ( --2*,3*/2 ) ; at ( pqautha.north ) @xmath131 ; ( pqauthb ) at ( + 2*,3*/2 ) ; at ( pqauthb.north ) @xmath132 ; ( pkeya ) at ( --3*,3*/2 ) ; ( keylabela ) at ( pkeya.north ) @xmath366 ; ( pkeyb ) at ( + 3*,3*/2 ) ; ( keylabelb ) at ( pkeyb.north ) @xmath367 ; ( compa ) ; at ( compa.north west ) composed protocol @xmath352 ; ( compb ) ; at ( compb.north west ) composed protocol @xmath353 ; ( aliceup ) at ( -,5*/2 ) ; ( alicemiddle ) at ( -,3*/2 ) ; ( alicedown ) at ( -,/2 ) ; ( bobup ) at ( , 5*/2 ) ; ( bobmiddle ) at ( , 3*/2 ) ; ( bobdown ) at ( , /2 ) ; ( eveleftl ) at ( --,- ) ; ( eveleftr ) at ( -+,- ) ; ( everightl ) at ( -,- ) ; ( everightr ) at ( + , - ) ; ( aq ) at ( --2*,3 * ) ; ( bq ) at ( + 2*,3 * ) ; ( ak1 ) at ( -+,3 * ) ; ( bk1 ) at ( -,3 * ) ; ( ac ) at ( --,2 * ) ; ( bc ) at ( + , 2 * ) ; ( ak2 ) at ( -+,2 * ) ; ( bk2 ) at ( -,2 * ) ; ( aliceup.center ) to node[auto , pos=.4 ] req .
( pkeya.west |- aliceup ) ; ( pkeya.west |- alicemiddle ) to node[auto , swap , pos=.6 ] @xmath17 ( alicemiddle.center ) ; ( alicedown.center ) to node[auto , pos=.4 ] @xmath121 ( pkeya.west |- alicedown ) ; ( bobup.center ) to node[auto , pos=.4,swap ] req .
( pkeyb.east |- aliceup ) ; ( pkeyb.east |- alicemiddle ) to node[auto , pos=.6 ] @xmath17 ( bobmiddle.center ) ; ( pkeyb.east |- alicedown ) to node[auto , pos=.6 ] @xmath121 ( bobdown.center ) ; ( pkeya.east |- aliceup ) to ( pqautha.west |- aliceup ) ; ( pqautha.west |- alicemiddle ) to ( pkeya.east |- alicemiddle ) ; ( pkeya.east |- alicedown ) to ( pqautha.west |- alicedown ) ; ( pkeyb.west |- aliceup ) to ( pqauthb.east alicemiddle ) ; ( pqauthb.east |- alicedown ) to ( pkeyb.west |- alicedown ) ; ( ak1 ) to ( aq ) ; ( aq ) to ( ak1 ) ; ( bk1 ) to ( bq ) ; ( bq ) to ( bk1 ) ; ( pcautha ) to ( pqautha ) ; ( pqauthb ) to ( pcauthb ) ; ( pqautha.east |- chforward ) to ( peccforwarda ) ; ( peccforwardb ) to ( pqauthb.west |- chforward ) ; ( ak2 ) to ( ac ) ; ( ac ) to ( ak2 ) ; ( bk2 ) to ( bc ) ; ( bc ) to ( bk2 ) ; ( peccbacka ) to ( pcautha.east |- chback ) ; ( pcauthb.west |- chback ) to ( peccbackb ) ; ( peccforwarda ) to ( eveleftl |- chforward ) to ( eveleftl.center ) ; ( eveleftr.center ) to ( eveleftr |- chforward ) to ( peccforwardb ) ; ( peccbackb ) to ( everightr |- chback ) to ( everightr.center ) ; ( everightl.center ) to ( everightl |- chback ) to ( peccbacka ) ; in the following we assume for simplicity that @xmath368 and take @xmath369 . plugging in @xmath370 and the parameters from the two explicit quantum authentication protocols from
, we get the following two corollaries . uses the explicit construction proposed by barnum et al .
@xcite . uses the explicit unitary @xmath9/design construction from chau@xcite ( see for details ) .
[ cor : construction.1 ] for any @xmath5 and @xmath7 , there exist an explicit protocol that requires @xmath371 bits of secret key , a forward noisy insecure quantum channel and a backward noisy insecure classical channel , to construct a secure channel for an @xmath5 qubit quantum message , which recycles all the key if the message is accepted and @xmath7 bits if the message is rejected , and has error @xmath372 follows from the composition theorem from ac @xcite as well as , , , , and .
[ cor : construction.2 ] for any @xmath5 and @xmath7 , there exists an explicit protocol that requires @xmath373 bits of secret key , a forward noisy insecure quantum channel and a backward noisy insecure classical channel , to construct a secure channel for an @xmath5 qubit quantum message , which recycles all the key if the message is accepted and @xmath374 bits if the message is rejected , and has error @xmath375 follows from the composition theorem from ac @xcite as well as , , , , and .
note that both schemes have a loss of @xmath376 key bits if the message is not successfully authenticated : @xmath142 bits used to one - time pad the quantum message of length @xmath5 , @xmath377 bits to one - time pad the two classical strings of @xmath7 bits
the @xmath7 bits of key sent to replace those consumed by the backwards authentic channel and the @xmath7 bits used as syndrome and finally the @xmath7 bits consumed by the backwards authentic channel can not be replaced , so they are lost as well .
the backwards authentic channel is crucial in the construction of the secure channel with key recycling analyzed in this section , because without it , alice would not know that her message arrived let alone whether it was accepted or not and thus not be able to recycle the key .
one can however skip the authentic channel if the players share a stronger resource , e.g. , another secure quantum channel that recycles key .
this is the case if the players construct secure quantum channels in both directions and alternate between the two : first alice sends a quantum message to bob , then bob to alice , then alice to bob , etc .
let them share two sets of keys , one set is used for the forward communication , the other is used for the backward communication .
if at any point a message is not successfully authenticated , this means that an eavesdropper is disturbing the communication , and the players abort and stop communicating .
thus , bob only sends his next message to alice if he successfully received her message .
so if alice successfully authenticates bob s message , she knows that hers was received , and can recycle all of her key to send the next message .
we thus avoid any explicit confirmation of reception , since sending the next message is in itself the confirmation . one case
easily show that the composition of @xmath7 rounds of this protocol , each round sending an @xmath5 qubit quantum message , constructs a secure channel for @xmath378 qubits , which we denote @xmath379 .
unlike the resource @xmath380 which gave the adversary a @xmath81 bit input to decide if the message is delivered or not , @xmath379 provides the adversary with a bit of input for every block of @xmath5 qubits but if one block is prevented from being delivered , none of the subsequent messages are delivered either .
the error of this construction is @xmath7 times the error of each round , @xmath381
. this may continue arbitrary long if no adversary introduces noise on the channel , since the quantum authentication protocol recycles every bit of key .
the family of quantum authentication protocols of barnum et al .
@xcite as well as the subset analyzed in this work are large classes , which include many protocols appearing independently in the literature .
the signed polynomial code @xcite , the clifford code @xcite ( which is a unitary @xmath117/design @xcite ) and the unitary @xmath382/design scheme from @xcite and all instances which use a strong purity testing code .
our results apply directly to the clifford and unitary @xmath382/design schemes which have in the same error as the unitary @xmath9/design scheme from .
but the signed polynomial code uses an ecc on qudits , not qubits , so our proof does not cover this case , and would have to be adapted to do so .
the trap code @xcite is an example of a quantum authentication scheme that uses a purity testing code that is not a strong purity testing code , i.e. , errors which do not modify the message do not necessarily provoke an abort .
for example , if the adversary performs a simple bit flip in one position , this will provoke an abort with probability @xmath383 in the variant from @xcite and with probability @xmath384 in the variant from @xcite , but leaves the message unmodified if no abort occurs .
if the adversary learns whether bob accepted the message or not , she will learn whether the ecc used detects that specific bit flip or not , and thus learn something about the key used to select the ecc .
hence , the players can not recycle the entire key , even in the case where the message is accepted .
the restriction to strong purity testing codes is thus necessary to recycle every bit .
it remains open how many bits of key can be recycled with the trap code . another quantum authentication scheme , auth - qft - auth , has been proposed in @xcite , where the authors prove that some of the key can be recycled as well .
we do not know if this scheme fits in the family from @xcite or not .
in the classical case , almost strongly @xmath9/universal hash functions @xcite are used for authentication , and any new family of such functions immediately yields a new mac .
likewise , any new purity testing code provides a new quantum authentication scheme .
however , it is unknown whether all quantum authentication schemes can be modeled as a combination of a one - time pad and a purity testing code , or whether there exist interesting schemes following a different pattern .
we have proven that a loss of @xmath12 bits of key is inevitable with these schemes if the adversary tampers with the channel . in the case of the unitary @xmath9/design scheme , which has the smallest error ,
this is @xmath385 bits of key which are consumed .
a loss of @xmath142 bits will always occur , since these are required to one - time pad the message .
it remains open whether there exist other schemes which do not fit the one - time pad + purity testing code model which recycle more key .
another open question is whether it is possible to find a prepare / and / measure scheme to encrypt and authenticate a classical message in a quantum state , so that all of the key may be recycled if it is successfully authenticated .
note that such a scheme would necessarily have to be designed to work on a noisy channel , because quantum eccs are not prepare / and / measure , so one can not compose it with an ecc to withstand a noisy channel as in this work .
some ideas for achieving this are discussed in @xcite .
cp would like to thank anne broadbent , frdric dupuis and debbie leung for useful discussions .
cp is supported by the european commission fp7 project raquel ( grant no .
323970 ) , us air force office of scientific research ( afosr ) via grant fa9550 - 16 - 1 - 0245 , the swiss national science foundation ( via the national centre of competence in research ` quantum science and technology ' ) and the european research council erc ( grant no .
258932 ) . [ app ]
as already mentioned in , the ac framework @xcite models cryptography as a resource theory . the ac framework does however not explicitly define these resources .
it follows a top / down paradigm and only specifies on each level of abstraction the properties of objects that are absolutely essential the axioms these objects must satisfy .
this simplifies the framework by removing unnecessary and cumbersome information
e.g. , a model of computation and results in more general framework that is not hard / coded with a specific communication or scheduling model . in this section
we give a brief introduction to ac . we illustrate this with an example taken from @xcite , namely that appending a message authentication code ( mac ) to a classical message is sufficient to construct a classical authentic channel given an insecure channel and a shared secret key .
we refer the interested reader to the original ac paper @xcite for a detailed treatment of the abstract layer and to @xcite for more gentle introductions to ac .
an @xmath386/_resource _ is an ( abstract ) system with interfaces specified by a set @xmath386 , e.g. , @xmath387 .
each interface @xmath388 is accessible to a user @xmath19 .
the objects depicted in are examples of resources .
the insecure channel in allows alice to input a message at her interface on the left and allows bob to receive a message at his interface on the right .
eve can intercept alice s message and insert a message of her choosing at her interface .
the authentic channel resource depicted in also allows alice to send a message and bob to receive a message , but eve s interface is more limited than for the insecure channel : she can only decide if bob receives the message or not , but not tamper with the message being sent .
the key resource drawn in provides each player with a secret key when requested .
additionally , resources are equipped with a parallel composition operator , @xmath389 , that maps two resources to another resource .
this is to be understood as both resources being merged into one , that can be accessed in any arbitrary order .
for example , if players share a secret key resource @xmath16 and a channel resource @xmath13 , they have the resource @xmath18 available , as depicted in .
given access to @xmath18 , the players could , e.g. , get a key from @xmath16 and use it to encode a message that is sent on @xmath13 _ converters _ capture operations that a player might perform locally at her interface .
these are ( abstract ) systems with two interfaces , an _ inside _ interface and an _ outside _ interface .
the inside interface connects to an interface of a resource , and the outside interface becomes the new interface of the resource resulting from the connection of this converter and resource .
for example , in the setting described a paragraph higher , alice might decide to append a mac to her message .
this is modeled as a converter @xmath20 that obtains the message @xmath21 at the outside interface , obtains a key at the inside interface from a key resource @xmath16 and sends @xmath22 on the insecure channel @xmath13 , where @xmath23 is taken from a family of strongly @xmath9/universal hash functions @xcite .
we illustrate this in .
converters are always drawn with rounded corners . if a converter @xmath24 is connected to the @xmath19 interface of a resource @xmath25 , we write @xmath26 or @xmath27 for the new resource obtained by connecting the two in this work we adopt the convention of writing converters at the @xmath50 and @xmath28 interfaces on the left and converters at the @xmath29 interface on the right , though there is no mathematical difference between @xmath26 and @xmath27
. serial and parallel composition of converters is defined as follows : @xmath390 by definition , converters at different interfaces commute , i.e. , @xmath391 if @xmath392 .
this allows us to draw systems as in without having to specify an order in which @xmath20 and @xmath365 are connected to the resource @xmath18 .
a protocol is then defined by a set of converters , one for every honest player .
another type of converter that we need is a _
filter_. the resources illustrated in depict a setting with an adversary that has some control over these resources . for a cryptographic protocol to be useful it is not sufficient to provide guarantees on what happens when an adversary is present , one also has to provide a guarantee on what happens when no adversary is present , e.g. , if no adversary tampers with the message on the insecure channel , then bob will receive the message that alice sent .
we model this setting by covering the adversarial interface with a filter that emulates an honest behavior./interfaces that are blank if the adversary is not active . ] in we draw an insecure and an authentic channel with filters @xmath39 and @xmath40 that transmit the message to bob . in the case of the insecure channel , one may want to model an honest noisy channel when no adversary is present .
this is done by having the filter @xmath39 add some noise to the message .
a dishonest player removes this and has access to a noiseless channel as in .
we use the term _ filtered resource _ to refer to a pair of a resource @xmath25 and a filter @xmath39 , and often write @xmath42 .
such an object can be thought of as having two modes : it is characterized by the resource @xmath43 when no adversary is present and by the resource @xmath25 when the adversary is present .
parallel composition of filtered resources naturally follows from parallel composition of resources and converters : @xmath393 the final object that is required by the ac framework to define the notion of construction and prove that it is composable , is a ( pseudo/)metric defined on the space of resources that measures how close two resources are . in the following , we use a distinguisher based metric , i.e. , the maximum advantage a distinguisher has in guessing whether it is interacting with resource @xmath25 or @xmath44 , which we write @xmath45 .
this is discussed further in .
we are now ready to define the security of a cryptographic protocol .
we do so in the three player setting , for honest alice and bob , and dishonest eve .
thus , in the following , all resources have three interfaces , denoted @xmath50 , @xmath28 and @xmath29 , and a protocol is then given by a pair of converters @xmath51 for the honest players .
we refer to @xcite for the general case , when arbitrary players can be dishonest . for convenience , we reproduce here .
[ def : security.app ] let @xmath52 be a protocol and @xmath53 and @xmath54 denote two filtered resources .
we say that _ @xmath55 constructs @xmath56 from @xmath57 within @xmath0 _ , which we write @xmath58 ,
if the two following conditions hold : 1 .
[ eq : def.cor.app ] we have @xmath59 2 . [ eq : def.sec.app ] there exists a converter @xmath60 which we call simulator such that @xmath61 if it is clear from the context what filtered resources @xmath57 and @xmath62 are meant , we simply say that @xmath55 is @xmath0/secure .
the first of these two conditions measures how close the constructed resource is to the ideal resource in the case where no malicious player is intervening , which is often called _ correctness _ in the literature .
the second condition captures _ security _ in the presence of an adversary .
for example , to prove that the mac protocol constructs an authentic channel from an insecure channel and a secret key within @xmath0 , we need to find a converter @xmath63 such that the real system @xmath64 can not be distinguished from the ideal system @xmath65 with advantage greater than @xmath0 . the simulator @xmath63 drawn in satisfies this requirement if the family of hash functions @xmath66 is @xmath0/almost strongly @xmath9/universal @xcite .
it follows from the composition theorem of the ac framework @xcite that if two protocols @xmath394 and @xmath395 are @xmath0- and @xmath396/secure , the composition of the two is @xmath397/secure . more precisely , let protocols @xmath394 and @xmath395 construct @xmath62 from @xmath57 and @xmath398 from @xmath62 within @xmath0 and @xmath396 , respectively , i.e. , @xmath399 it then follows from the triangle inequality of the metric that @xmath400 constructs @xmath398 from @xmath57 within @xmath401 , @xmath402 a similarly statement holds for parallel composition .
let @xmath394 and @xmath395 construct @xmath62 and @xmath403 from @xmath57 and @xmath404 within @xmath0 and @xmath396 , respectively , i.e. , @xmath405 if these resources and protocols are composed in parallel , we find that @xmath406 constructs @xmath407 from @xmath408 within @xmath401 , @xmath409 proofs of these statements can be found in @xcite .
as stated at the beginning of this section , the ac framework @xcite specifies only the necessary axioms that resources and converters must satisfy so that one can prove that the resulting notion of construction is composable .
modeling concrete systems such as those in figures [ fig : channel.dishonest ] , [ fig : classicalauth.real ] , [ fig : channel.honest ] or [ fig : classicalauth.ideal ] , requires an instantiation of the framework with mathematical objects that capture interactive quantum information / processing systems .
such an instantiation has been given in @xcite and proven to satisfy the axioms of ac , where the interactive systems are called _
causal boxes_. unlike the model of systems used in quantum uc @xcite , in which the output of a system is given by a quantum message and a classical string denoting a recipient , causal boxes allow messages to be sent to a superposition of different players ; they even allow superpositions of different numbers of messages to be generated in superpositions of different orders @xcite .
this generality is however not needed in the current work , because all converters and resources involved in the construction of secure channels have a very simple structure . for this work
, a system @xmath44 can be modeld as having internal memory @xmath410 , and sets of in - ports @xmath411 and out - ports @xmath412 with message spaces @xmath413 and @xmath414 , respectively . furthermore , upon receiving a message at @xmath415 , @xmath44 always produces outputs at a fixed set of out - ports @xmath416 such that @xmath417 if @xmath392 .
a system @xmath44 is thus entirely described by a set of completely positive , trace / preserving ( cptp ) maps @xmath418 where @xmath419 and @xmath420 is the space of linear operators on @xmath421 . upon receiving a message at some port @xmath415 , the system
@xmath44 then applies the map @xmath422 and outputs the messages in the registers @xmath423 . for a fixed order of messages received ,
this specific type of system has been called a _ quantum strategy _
@xcite , _ quantum comb _
@xcite or _ operator tensor _
@xcite here we use the terminology from @xcite , namely _
comb_. a comb can be represented more compactly as a single cptp map @xmath424 , and using the composition rules for combs @xcite or causal boxes @xcite , two such systems can be composed to obtain a new system of the same type . the exact formula for composing systems is not needed in the current work ;
in all special cases where we connect two systems , the resulting system can easily be worked out by hand .
we refer the interested reader to @xcite for the generic cases .
as mentioned at the end of we use a distinguisher metric to define the distance between two systems @xmath25 and @xmath44 .
this means that another system @xmath46 , a _ distinguisher _ , is given access to either @xmath25 or @xmath44 , and has to guess to which of the two it is connected .
let @xmath47 $ ] and @xmath48 $ ] be the binary random variables corresponding to @xmath46 s guess , then the distinguishing advantage between @xmath25 and @xmath44 for this specific distinguisher is given by @xmath425 = 0 \right ] - \pr\left [ \ad[\as ] = 0 \right ] \right|\,,\ ] ] and the distance is given by @xmath426 where the supremum is taken over all distinguishers allowed by quantum mechanics
it has been proven in @xcite that @xmath427 is indeed a metric . due to the simple structure of the systems considered in this work , the distinguishing metric can be reduced to the following strategy for the general case of the distinguishing metric between causal boxes we refer the reader to @xcite .
let the distinguisher have internal memory @xmath428 .
it choses an in - port @xmath429 , and prepares a state @xmath430 .
the @xmath431 part is sent to the system at the @xmath429 port .
it then receives the output on ports @xmath432 , which it appends to its internal memory .
it measures its internal memory to decide on the next in - port @xmath433 , applies a map @xmath434 , and inputs the @xmath435 part at the corresponding port .
this process is repeated until there are no more unused ports , after which it measures its internal memory and produces one bit of output , its final guess . in the case
where the two systems being compared only have one in - port , this metric reduces to the diamond norm . and
if the systems have no in - port ( or one trivial in - port of dimension @xmath81 ) , this results in the trace / distance between the states output by the two systems .
the concept of a unitary @xmath9-design was originally proposed in @xcite .
the following ( equivalent ) definition is taken from @xcite .
a unitary @xmath9-design is a finite set of unitaries @xmath436 on @xmath437 such that for all @xmath438 @xmath439 where @xmath440 is the haar measure .
equivalently , @xmath441 is a unitary @xmath9-design if for any quantum channel @xmath442 and state @xmath443 , @xmath444 one way to construct a unitary @xmath9-design is by finding a set of unitaries @xmath111 that map all non / identity paulis to each other with equal frequency , i.e. , @xmath445 such that @xmath297 and @xmath446 , @xmath447 where @xmath448 is some global phase and @xmath294 is the dimension of the hilbert space .
a unitary @xmath9/design is then obtained by composing these unitaries with paulis , i.e. , the set @xmath449 is a unitary @xmath9/design .
this has been used in @xcite to show that the clifford group is a unitary @xmath9-design , and is further discussed in @xcite .
chau @xcite finds a set @xmath111 satisfying . to understand his construction
, we must view the indices @xmath21 and @xmath74 of a pauli operator @xmath72 as elements of a galois field @xmath450 .
let @xmath451 be any element of the special linear group of @xmath452 matrices over the finite field @xmath453 , i.e. , matrices with determinant @xmath81 .
chau then finds unitaries @xmath454 such that @xmath455 for some global phase @xmath456 that may depend on @xmath457 , where the arithmetic in the indices is done in @xmath453 . by considering the entire set
@xmath458 one can verify that is satisfied .
since @xmath459 , we need @xmath460 bits of key to chose the unitary .
we now show that this set is a strong purity testing code .
[ lem : purity ] any set @xmath111 satisfying with @xmath461 is a strong purity testing code with @xmath300 .
an error @xmath73 is not detected if it is mapped to @xmath462 .
there are @xmath463 such paulis @xmath98 which are not identity .
since the unitaries @xmath464 are constructed to map @xmath73 to all non / identity @xmath98 with equal frequency , then @xmath465 of them will not detect @xmath73 .
a unitary @xmath305/design is defined similarly to a unitary @xmath9/design , except that it has a @xmath305/fold tensor product instead of a @xmath9/fold tensor product in . intuitively , if a unitary @xmath9/design is a good quantum authentication scheme , then so should any ( approximate ) unitary @xmath305/design for any @xmath466 .
the argument we use in the proof is that a unitary @xmath9/design satisfies .
this also holds for unitary @xmath117/designs ( given by the clifford group @xcite ) , but not necessarily for @xmath467 , where the unitaries are not elements of the clifford group anymore .
one can however show directly from that a unitary @xmath9/design is a strong purity testing code,/design scheme , is that chau s construction @xcite is a subset of a unitary @xmath9/design that satisfies but not .
we still have to compose it with a random pauli to obtain the unitary @xmath9/design . ] and since all ( approximate ) unitary @xmath305/designs ( approximately ) satisfy , they can all be used to construct quantum authentication schemes that have the same key recycling properties as unitary @xmath9/designs .
[ lem : purity.2 ] any @xmath1/approximate @xmath305/design with @xmath466 is a strong purity testing code with error @xmath468 . to prove that a set of unitaries is a strong purity testing code , one has to show that it can be used to detect all non / identity pauli errors with high probability .
setting @xmath469 for a non / identity pauli @xmath73 in , one can show that that the rhs becomes ( see , e.g. , @xcite ) , @xmath470 if @xmath461 , @xmath471 where @xmath345 is the first @xmath5 qubits of @xmath121 and the last @xmath7 qubits are used as syndrome , then the probability of obtaining @xmath41 when performing a measuring in the computational basis on the syndrome is @xmath472 thus , if the distance between the lhs and rhs of is @xmath1 , then the probability of not detecting a pauli error is less than @xmath468 .
[ lem : trace2norm ] let @xmath92 and @xmath473 be two subnormalized states .
then @xmath474 where @xmath224 is the vector @xmath9/norm .
it was shown in ( * ? ? ?
* lemma a.2.3 ) that if @xmath475 is real , then @xmath476 for complex @xmath475 we define @xmath477 .
it then follows from ( * ? ? ?
* lemma a.2.3 ) that @xmath478 to prove this lemma it remains to show that @xmath479 since combining this with @xmath480 we get .
dorit aharonov , michael ben - or , and elad eban .
interactive proofs for quantum computations . in _ proceedings of innovations in computer science , ics 2010 _ , pages 453469 .
tsinghua university press , 2010 . .
charles h. bennett and gilles brassard .
quantum cryptography : public key distribution and coin tossing . in _ proceedings of ieee international conference on computers , systems , and
signal processing _ , pages 175179 , 1984 .
howard barnum , claude crpeau , daniel gottesman , adam smith , and alain tapp .
authentication of quantum messages . in
_ proceedings of the 43rd symposium on foundations of computer science , focs 02 _ , pages 449458 .
ieee , 2002 . .
michael ben - or , claude crpeau , daniel gottesman , avinatan hassidim , and adam smith .
secure multiparty quantum computation with ( only ) a strict honest majority . in _ proceedings of the 47th symposium on foundations of computer science , focs 06 _ , pages 249260 , 2006 . .
anne broadbent , gus gutoski , and douglas stebila .
quantum one - time programs . in _ advances in cryptology
crypto 2013 _ , volume 8043 of _ lecture notes in computer science _ , pages 344360 .
springer , 2013 .
. ran canetti .
universally composable security : a new paradigm for cryptographic protocols . in _ proceedings of the 42nd symposium on foundations of computer science , focs 01 _ , pages 136145 .
ieee , 2001 . .
frdric dupuis , jesper buus nielsen , and louis salvail .
actively secure two - party evaluation of any quantum operation . in reihaneh safavi - naini and ran canetti , editors , _ advances in cryptology
crypto 2012 _ , volume 7417 of _ lecture notes in computer science _ , pages 794811 .
springer , 2012 . .
ivan damgrd , thomas brochmann pedersen , and louis salvail . a quantum cipher with near optimal key - recycling . in _ advances in cryptology
crypto 2005 _ , volume 3621 of _ lecture notes in computer science _ ,
pages 494510 .
springer , 2005 . .
ueli maurer .
constructive cryptography a new paradigm for security definitions and proofs . in _ proceedings of theory of security and applications , tosca 2011 _ , volume 6993 of _ lecture notes in computer science _ , pages 3356 .
springer , 2012 . .
renato renner and robert knig .
universally composable privacy amplification against quantum adversaries . in joe
kilian , editor , _ theory of cryptography , proceedings of tcc 2005 _ , volume 3378 of _ lecture notes in computer science _
, pages 407425 .
springer , 2005 . .
valerio scarani , helle bechmann - pasquinucci , nicolas j. cerf , miloslav du ek , norbert ltkenhaus , and momtchil peev .
the security of practical quantum key distribution .
, 81:13011350 , september 2009 . .
dominique unruh .
universally composable quantum multi - party computation . in _ advances in cryptology
eurocrypt 2010 _ , volume 6110 of _ lecture notes in computer science _ , pages 486505 .
springer , 2010 . .
|
we show that a family of quantum authentication protocols introduced in [ barnum et al . ,
focs 2002 ] can be used to construct a secure quantum channel and additionally recycle all of the secret key if the message is successfully authenticated , and recycle part of the key if tampering is detected .
we give a full security proof that constructs the secure channel given only insecure noisy channels and a shared secret key .
we also prove that the number of recycled key bits is optimal for this family of protocols , i.e. , there exists an adversarial strategy to obtain all non - recycled bits .
previous works recycled less key and only gave partial security proofs , since they did not consider all possible distinguishers ( environments ) that may be used to distinguish the real setting from the ideal secure quantum channel .
|
solar energetic particle ( sep ) events are major space weather phenomena associated with explosive processes occurring in the solar atmosphere , such solar flares and coronal mass ejections ( cmes ) .
seps can produce hazardous effects to manned and robotic flight missions in the near - earth space environment , and influence the atmospheric chemistry and dynamics .
large sep events can strongly perturb the earth s magnetic field , inducing geomagnetic storms and modifying the cosmic - ray ( cr ) access to the inner magnetosphere .
the consequent reduction in the geomagnetic shielding can significantly increase the potential radiation exposure compared with geomagnetically quiet times .
estimates of geomagnetic cutoffs have been provided by satellite observations and theoretical calculations @xcite mainly based on tracing particles through models of the earth s magnetic field @xcite . in this work we present pamela s measurements of the variability of the geomagnetic cutoff during the sep events on 2006 december 13 and 14 , with the focus on the strong magnetic storm on december 14 and 15 .
pamela is a space - based experiment designed for a precise measurement of the charged cosmic radiation in the kinetic energy range from some tens of mev up to several hundreds of gev @xcite . in particular
, pamela is providing accurate measurements of seps in a wide energy interval @xcite , bridging the low energy data by other spacecrafts and the gle data by the worldwide network of neutron monitors ; in addition , the detector is sensitive to the particle composition and is able to reconstruct flux angular distributions @xcite , enabling a more complete view of sep events .
the resurs - dk1 satellite , which hosts the apparatus , was launched into a semi - polar ( 70 deg inclination ) and elliptical ( 350@xmath2610 km altitude ) orbit on 2006 june 15 .
the spacecraft is 3-axis stabilized ; its orientation is calculated by an onboard processor with an accuracy better than 1 deg .
particle directions are measured with a high angular resolution ( @xmath3 2 deg ) .
details about apparatus performance , proton selection , detector efficiencies and experimental uncertainties can be found elsewhere ( e.g. @xcite ) .
the selected data set includes protons acquired by pamela between 2006 december 12 and 18 .
the analysis described in this work is based on the igrf-11 @xcite and the ts05 @xcite models for the description of the internal and external geomagnetic field , respectively .
the ts05 model is a high resolution dynamical model of the storm - time geomagnetic field , based on recent satellite measurements ; consistent with the data - set coverage , it is valid for @xmath4 @xmath5 -15 earth s radii ( re ) . for comparison purposes ,
the t96 model @xcite ( valid up to 40 re ) was used as well .
solar wind ( sw ) and interplanetary magnetic field ( imf ) parameters were obtained from the high resolution ( 5-min ) omniweb database @xcite .
data were analyzed in terms of altitude adjusted corrected geomagnetic ( aacgm ) coordinates , developed to provide a more realistic description of high latitude regions by accounting for the multipolar geomagnetic field .
they are defined such that all points along a magnetic field line have the same geomagnetic latitude and longitude , so that they are closely related to invariant magnetic coordinates @xcite .
the aacgm reference frame coincides with the standard corrected geomagnetic ( cgm ) coordinate system @xcite at the earth s surface . unlike other commonly used variables such as the invariant latitude
, the computation of such coordinates at low earth orbits is not significantly affected by the modeling of external geomagnetic sources .
the lowest magnetic latitude to which a cr particle can penetrate the earth s magnetic field is known as its _ cutoff latitude _ and is a function of the particle momentum per unit charge , which is referred to as its rigidity .
alternatively one may consider a _
cutoff rigidity _ corresponding to a given location in space , i.e. the minimum rigidity needed to access to that location .
some complications arise from the presence of the earth s solid body ( together with its atmosphere ) : both `` allowed '' and `` forbidden '' bands of cr particle access are present in the so - called `` penumbra '' region @xcite .
the numerical algorithm developed to extract cutoff latitudes from the pamela data is similar to one used by @xcite and @xcite .
for each rigidity bin , a mean flux was obtained by averaging fluxes above 65 degrees latitude , and the cutoff latitude was evaluated as the latitude where the flux intensity is equal to the half of the average value .
alternatively , cutoff latitudes were estimated with back - tracing techniques @xcite .
using the spacecraft ephemeris data , and the particle rigidity and direction provided by the pamela tracking system , trajectories of all detected protons were reconstructed by means of a tracing program based on numerical integration methods @xcite , and implementing the afore - mentioned geomagnetic field models .
trajectories were back propagated from the measurement location until they escaped the model magnetosphere boundaries ( solar or galactic crs ) or they reached an altitude of 40 km ( re - entrant albedo crs ) . at a given rigidity ,
the cutoff latitude was evaluated as the latitude where an equal percentage of interplanetary and albedo crs was registered .
the calculation was performed for 13 rigidity logarithmic bins , covering the interval [email protected] gv .
accounting for the limited statistics at highest rigidities , final cutoff values were derived by fitting averaged pamela observations over single orbital periods ( @xmath194 min ) .
on 2006 december 13 at 02:14 ut , an x3.4/4b solar flare occurred in the active region noaa 10930 ( s06w23 ; noaa - stp 2006 ) .
this event also produced a full - halo cme with the sky plane projected speed of 1774 km s@xmath6 .
the forward shock of the cme reached earth at about 14:10 ut on december 14 , causing a forbush decrease of galactic cr intensities that lasted for several days .
such large events are untypical of the intervals of low solar activity .
the flare x1.5 ( s06w46 ) at 21:07 ut on december 14 gave start to a new growth of particle intensity as recorded by pamela and other satellites .
the maximum energy of protons was below 1 gev , and therefore no ground level enhancement ( gle ) was recorded .
the corresponding cme had a velocity of 1042 km s@xmath6 .
pamela s measurements of the 2006 december sep fluxes can be found in publications @xcite .
figure [ corrplotimf_pam_rsel0_ir20_iaacgl20_k1 ] reports the variations in the imf ( bx , by and bz components in the gsm frame ) and sw ( dynamic pressure , velocity and density ) variables between 2006 december 12@xmath218 .
the large increase in the sw velocity associated with the leading edge of the cme caused a sudden commencement of a geomagnetic storm .
the initial phase of the storm , lasting up to about 23:00 ut , was characterized by intense fluctuations in the sw density and in all imf components . at a later stage ,
the imf bz component became negative , the sw density decreased , and the main phase of the storm started , reaching a maximum between 02:00@xmath208:00 ut on december 15 . another interplanetary shock associated with a different cme was observed on december 16 .
figure [ ctimevarfit_pam_aacglat_ir20_iaacgl20_k2 ] shows the geomagnetic cutoff latitudes measured by pamela for different rigidity bins ( color code ) .
each point denotes the cutoff latitude value averaged over a single spacecraft orbit ; the error bars include the statistical uncertainties of the measurement .
data were missed from 10:00 ut on december 13 until 09:14 ut on december 14 because of an onboard system reset of the satellite .
the evolution of the magnetic storm of december 14 and 15 followed the typical scenario in which the cutoff latitudes move equatorward as a consequence of a cme impact on the magnetosphere with an associated transition to southward bz . the registered cutoff variation decreases with increasing rigidity , with a maximum suppression ranging from about 6 deg in the lowest rigidity bin ( @xmath180 mev energy ) to about 2 deg in the highest rigidity bin ( @xmath13 gev energy ) .
figure [ comparison ] reports the comparison between measured and modeled cutoff latitudes , for the lowest rigidity bin [email protected] gv . while the t96 model appears to underestimate ( up to 4% ) the observations , a much better agreement can be noted between pamela and ts05 results .
however , the ts05 cutoff latitudes overestimate ( up to 2% ) the pamela ones during the storm main phase .
finally , figure [ corrplotgm_pam_rsel0_ir20_iaacgl20_k2 ] demonstrates the utility of the three indices used to infer cutoff latitude : the magnetic activity index ( kp ) , the disturbance storm time index ( dst ) and the sym - h index , measured using ground - based magnetometers , at 3-hour , 1-hour , and 1-min resolutions , respectively .
in general , the shapes of the time variations in the cutoff measurements are well correlated with corresponding indexes changes ( corresponding correlation coefficients are 0.8 , 0.78 and 0.78 , respectively ) . a better agreement is observed for kp during the initial phase of the storm , while the dst and the sym - h indexes show an improved correlation during the main and the recovery phases .
in this study we have exploited the data of the pamela satellite experiment to perform a measurement of the geomagnetic cutoff variations during the long lasting sep events of 2006 december 13 and 14 .
a significant reduction in the geomagnetic shielding was observed during the consequent strong magnetic storm on december 14 and 15 , with a maximum cutoff latitude suppression of about 6 deg for @xmath180 mev protons .
results were compared with those obtained with back - tracing techniques .
the observed cutoff variations are well correlated with the time profiles of the geomagnetic indexes ( kp , dst and sym - h ) . k. b. baker & s. wing , 1989 , j. geophys .
94 , pp 91399143 .
a. brekke , 1997 , physics of the upper polar atmosphere , wiley , new york , pp .
127145 , doi:10.1007/978 - 3 - 642 - 27401 - 5 .
a. bruno , et al . , 2015 ,
a. bruno , et al . , 2015 ,
pos(icrc2015)085 .
d. f. smart & m. a. shea , 1985 , j. geophys .
res . , 90 , 183 - 190 .
d. f. smart & m. a. shea , 2000 , final report , grant nag58009 , center for space plasmas and aeronomic research , the university of alabama in huntsville .
d. f. smart & m. a. shea , 2003 , adv .
space res .
, 32(1 ) , 103 - 108 .
d. f. smart & m. a. shea , 2005 , adv .
space res . , 36 , 20122020 .
n. a. tsyganenko , 1996 , in proc . of the third international conference on substorms ( ics-3 ) , versailles ,
france , 1217 may 1996 , edited by e. rolfe and b. kaldeich , eur .
space agency spec .
publ . , esa - sp , 389 , p. 181
|
data from the pamela satellite experiment were used to measure the geomagnetic cutoff for high - energy ( @xmath0 80 mev ) protons during the solar particle events on 2006 december 13 and 14 .
the variations of the cutoff latitude as a function of rigidity were studied on relatively short timescales , corresponding to single spacecraft orbits ( about 94 minutes ) . estimated
cutoff values were cross - checked with those obtained by means of a trajectory tracing approach based on dynamical empirical modeling of the earth s magnetosphere .
we find significant variations in the cutoff latitude , with a maximum suppression of about 6 deg for @xmath180 mev protons during the main phase of the storm .
the observed reduction in the geomagnetic shielding and its temporal evolution were compared with the changes in the magnetosphere configuration , investigating the role of imf , solar wind and geomagnetic ( kp , dst and sym - h indexes ) variables and their correlation with pamela cutoff results .
|
after 30 years since their detection by the vela satellites , we just now start to understand the physics of gamma ray bursts ( grb ) .
this has been made possible by the precise location of the wide field camera ( wfc ) of @xmath3sax , which allowed the detection of their x ray afterglow emission ( costa et al .
1997 ) and the optical follow up observations , leading to the discovery that they are cosmological sources ( van paradijs et al . 1997 ) .
the huge energy and power releases required by their cosmological distances support the fireball scenario ( cavallo & rees 1978 ; rees & mszros 1992 ; mszros & rees 1993 ) , even if we do not know yet which kind of progenitor makes the grb phenomenon .
the most accepted picture for the burst and afterglow emission is the internal / external shock scenario ( rees & mszros 1992 ; rees & mszros 1994 ; sari & piran 1997 ) . according to this scenario , the burst emission is due to collisions of pairs of relativistic shells ( internal shocks ) , while the afterglow is generated by the collisionless shocks produced by shells interacting with the interstellar medium ( external shocks ) .
the @xmath0ray light curves show an extremely variable emission , with variability timescales as short as a few milliseconds .
this , coupled with the huge powers involved , requires that the emitting plasma is moving at relativistic speeds , with bulk lorentz factors @xmath4 , to avoid strong suppression of high energy @xmath0rays due to photon photon collisions .
indeed , all the radiation we see is believed to come from the transformation of ordered kinetic energy of the fireball into random energy .
this must however happen at some distance from the explosion site , to allow the shells to be transparent to the produced radiation .
fiducial numbers for these distances are @xmath5@xmath6 cm for the internal shocks , and @xmath7 cm for external ones .
it is reasonable to assume that internal and external shocks can amplify seed magnetic fields and accelerate electrons to relativistic random energies . if this is the case , then the main radiation mechanism is the synchrotron process , responsible for both the @xmath0ray and for the afterglow emission .
there are strong evidences that this is the main process operating during the afterglow : the power law decay of the flux with time , the observed power law energy spectra ( for reviews see piran 1999 ; mszros 1999 ) , and the recently detected linear optical polarization in grb 990510 ( covino et al . 1999 ,
wijers et al .
1999 ) . for the bursts itself ,
the main evidence is the prediction of the typical energy at which the observed spectrum peaks , in a @xmath8 representation .
however , the very same synchrotron shock scenario ( iss ) inevitably predicts very fast radiative cooling of the emitting particles , with a resulting severe disagreement between the predicted and the observed spectrum ( ghisellini , celotti & lazzati 1999 , see also cohen et al .
1997 ; sari , piran & narayan 1998 ; chiang 1999 ) . in the following , after having briefly recalled some basic facts concerning grbs , i will discuss the problems with the synchrotron interpretation of the @xmath0ray emission and some possible alternatives .
another hot issue in the grb field is the possible collimation of their emitting plasma , leading to anisotropic emission able to relax the power requirements , at the expense of an increased burst event rate .
i will show that in this respect polarization studies could be crucial , since there can be a link between the deceleration of a _ collimated _ fireball , the time behavior of the polarized flux and its position angle , and the light curve of the total flux .
_ duration _ the duration distribution is clearly bimodal .
_ short _ burst durations range between 0.01 to 2 seconds .
_ long _ burst last from 2 to a few hundred seconds .
all information derived from the precise location of grbs refer to _ long _ bursts .
there is some indication that short bursts may have a different @xmath9 ( tavani 1998 )
. 0.2 true cm _ spectra and fluences _ most @xmath0ray fluences ( i.e. the flux integrated over the duration of the burst ) are in the range @xmath10@xmath11 erg @xmath12 .
grb spectra are very hard , and can be fitted by two smoothly connected power laws , implying a peak ( in a @xmath13@xmath8 plot ) at an energy @xmath14 of a few hundreds kev .
below @xmath14 the spectrum has a slope @xmath15 in the range [ @xmath161 , @xmath170.5 ] ( preece et al .
1998 , lloyd & petrosian 1999 ; @xmath18 ) .
this spectral index varies during the burst , as does @xmath14 .
the general trend is that the spectrum softens , and @xmath14 decreases , with time .
more precise statements must however wait for larger area detectors , since what we inevitably do , at present , is to fit a time integrated spectrum of a very rapidly variable source : the minimum integration time is @xmath191 second for the strongest bursts , while the variability timescales can be hundreds of times shorter .
some bursts have been detected at very large @xmath0rays energies ( @xmath20 100 mev ) ( see the review by fishman & meegan 1995 , and references therein ) .
0.2 true cm _ x ray afterglow
_ nearly all of the burst detected by the bepposax grbm and wfc , and that could be followed with the narrow field instruments of bepposax a few ( 610 ) hours after the event , showed an x ray afterglow , whose flux decays in time as a power law .
0.2 true cm _ optical afterglow _ for about 2/3 of the bursts with good locations an optical afterglow has been detected .
the monochromatic flux decreases in time as a power law @xmath21 , with @xmath22 in the range 0.82 .
remarkably , grb 990123 has been detected by the robotic telescope rotse 22 seconds after the @xmath0ray trigger at @xmath23 , reaching @xmath24 47 seconds after the trigger ( akerlof & mckay 1999 ) .
usually , the magnitudes of the optical afterglow detected a few hours from the event are in the range 1821 .
0.2 true cm _ radio afterglow
_ a few bursts showed radio emission , usually some time ( @xmath19days ) after the burst event .
violent radio flux variations in grb 970508 ( frail et al . 1997 ) and grb 980329 ( taylor et al . 1998 ) have been interpreted as due to interstellar scintillation , effective if the source is extremely compact .
transition from the phase of violent activity to a more quiescent phase has been interpreted as due to the expansion of the radio emitting source , beautifully confirming the relativistic expansion hypothesis .
0.2 true cm _ redshifts _ up to july 15 , 1999 , we know the redshift of 8 grbs , with an additional one ( grb 980329 ) estimated by a cut - off in the spectrum interpreted as @xmath25 absorption ( fruchter 1999a ) , and another one ( grb 980425 ) of controversial identification with the supernova sn1998bw . in table 1 we list some observed characteristics of the grbs of known redshifts , and fig .
1 shows the power emitted in @xmath0rays , the fluences and the magnitude of the host galaxy candidates as a function of redshift .
-0.5 true cm -0.5 true cm 0.3 true cm * tab .
1 - gamma ray burst of known redshift * -0.4 true cm [ cols="<,^,^,^,^,^,^",options="header " , ] -0.3 true cm _ notes : _ @xmath26 : batse @xmath0ray fluences in units of @xmath27 erg @xmath12 .
@xmath28 : wfc flux in crab units .
_ references _ : 1 : djorgovski et al . 1999b ; 2 :
metzger et al .
1997 ; 3 : kulkarni et al .
1998b 4 : fruchter 1999a ; 5 : tinney et al . 1998 ; 6 : djorgovski et al .
1999a ; 7 : djorgovski et al .
1998a ; 8 : kelson et al .
1999 ; 9 : hjorth et al . 1999 ; 10 : vreeswijk et al . 1999 ; 11 : galama et al . 1999
; 12 : fruchter et al . 1999b ; 13 : bloom et al . 1998a
; 14 : odewann et al .
1998 ; 15 : djorgovski et al .
1998a ; 16 : kulkarni et al . 1998b
; 17 : djorgovski et al .
1998c ; 18 : bloom et al . 1999b
; 19 : fruchter et al . 1999c ; 20 : fruchter et al . 1999d .
the energy involved in @xmath0ray burst explosions is huge .
no matter in which form the energy is initially injected , a quasi thermal equilibrium between matter and radiation is reached , with the formation of electron positron pairs accelerated to relativistic speeds by the high internal pressure .
this is a _
fireball_. when the temperature of the radiation ( as measured in the comoving frame ) drops below @xmath1950 kev the pairs annihilate faster than the rate at which are produced .
but the presence of even a small amount of barions , corresponding to only @xmath29 , makes the fireball opaque to thomson scattering : the internal radiation thus continues to accelerate the fireball until most of its initial energy has been converted into bulk motion . after this phase
the fireball expands at a constant speed and at some point becomes transparent .
if the central engine is not completely impulsive , but works intermittently , it can produce many shells ( i.e. many fireballs ) with slightly different lorentz factors .
late but faster shells can catch up early slower ones , producing shocks which give rise to the observed burst emission . in the meantime ,
all shells interact with the interstellar medium , and at some point the amount of swept up matter is large enough to decelerate the fireball and produce other radiation which can be identified with the afterglow emission observed at all frequencies .
in the comoving frame of the faster shell , protons of the other shell have an energy density @xmath30 , where @xmath31 is the bulk lorentz factor of the slower shell measured in the rest frame of the other .
the magnetic energy density @xmath32 can be amplified to values close to equipartition with the proton energy density , @xmath33 .
the proton density of the shell can be estimated by its kinetic power : @xmath34 , yielding @xmath35 .
also each electron can share a fraction of the available energy , and if there is one electron for each proton , namely _ if electron
positron pairs are not important _ , then @xmath36 .
these simple hypotheses lead to a predicted observed synchrotron frequency @xmath37 in very good agreement with observations .
note that the ` equipartition coefficients ' , @xmath38 and @xmath39 , must be close to unity for the observed value of @xmath14 to be recovered . in turn
this also implies / requires that pairs can not significantly contribute to the lepton density .
the synchrotron process is a very efficient radiation process . with the strong magnetic fields required to produce the observed @xmath0rays , the synchrotron cooling
time is therefore very short .
as pointed out by ghisellini , celotti & lazzati ( 1999 ) , the cooling time ( in the observer frame ) can be written as : @xmath40 where @xmath41 is the radiation energy density
. since the shortest integration times are of the order of 1 s , the observed spectrum is _ always _ the time integrated spectrum produced by a rapidly cooling particle distribution . since @xmath42 , in order to conserve the particle number , the instantaneous cooling distribution has to satisfy @xmath43 . when integrated over time , the contribution from particles with different lorentz factors is ` weighted ' by their cooling timescale @xmath44 .
therefore the predicted ( integrated ) flux spectrum is @xmath45 , extending from @xmath191 kev to @xmath46mev energies .
we thus conclude that , within the assumptions of the iss , a major problem arises in interpreting the observed spectra as synchrotron radiation .
ghisellini , celotti & lazzati ( 1999 ) have discussed possible ` escape routes ' , such as deviations from equipartition , fastly changing magnetic fields , strong cooling by adiabatic expansion and re acceleration of the emitting electrons .
none of these possibilities help .
the drawn conclusion is that the burst emission is probably produced by another radiation process .
an alternative is quasi thermal comptonization ( ghisellini & celotti 1999 ; liang 1997 , thompson 1994 ) .
why do grb spectra peak at a few hundreds kev ? this must be due to a quite robust mechanism and/or a feedback process . the fact that @xmath47 is close to @xmath48 is tempting .
i discuss below three possibilities .
assume that the main emission process is quasi thermal comptonization ( ghisellini & celotti 1999 ) by a distribution of electrons ( and positrons ) peaked at subrelativistic energies ( not necessarily thermal ) .
photons produced above the pair production threshold get absorbed and create pairs .
this process has been extensively studied in the past assuming thermal particles , steady state and pair equilibrium ( equal pair production and annihilation rates ) .
the main result is that for compact sources the maximum temperature is constrained in a narrow range [ 30300 kev ] ( svensson 1982 , 1984 ) : for increasing luminosities the equilibrium pair density increases , more particles are then sharing the available energy , making the temperature to decrease .
if an high energy tail is present , more photons are created above the threshold for pair production with respect to the case of a pure maxwellian , and thus pairs become important for temperatures smaller than in the completely thermal case ( see e.g. stern 1999 ) .
brainerd ( 1994 ) linked the typical high energy cut off of grbs to the effect of down scattering : photons with energies much larger than @xmath48 pass undisturbed through a scattering medium because of the reduction with energy of the klein nishina cross section , while photons with energies just below @xmath48 interact , and their energy after the scattering is reduced .
the net effect is to produce a downscattering hole " in the spectrum , between @xmath49 and @xmath50 .
the attractive feature of this model is that the cut off energy
is associated with @xmath48 .
the difficulty is that a significant part of the power originally radiated by the burst goes into heating ( by the compton process ) of the scattering electrons .
suppose that the circumburst material has a density @xmath51 and that , close to the burst emission site ( i.e. between @xmath52 and @xmath53 ) , the scattering optical depth is @xmath54 .
this material scatters back a fraction @xmath55 of the burst power . if we require that the primary spectrum is not modified by photon
photon absorption , the optical depth of the scattering matter and its density must be ( ghisellini & celotti 1999 ) @xmath56 this requirement is particularly severe , especially in the case of bursts originating in dense stellar forming regions . on the other hand , photon
photon opacity may be an important ingredient to shape the spectrum , and _ the _ reason why grb spectra peak at around 300 kev .
the pairs created ahead of the fireball radiate their energy in a short time , and are re accelerated by the incoming fireball .
the net effect may be simply to increase the density of the radiating particles , introducing a feedback process : an increased density lowers the effective temperature @xmath57 less energy is radiated above @xmath58 @xmath57 the number of pairs produced via the mirror " process decreases @xmath57 the new pair density decreases , and so on .
another feedback is introduced by the fact that the photons scattered back to the emitting shell will increase the number of seed photons and the radiative cooling rate , softening the spectrum .
covino et al . ( 1999 ) and wijers et al . ( 1999 ) detected a small , but highly significant , degree of linear polarization of the optical afterglow of grb 990510 : @xmath59% 18.5 and 20.5 hours after the burst , respectively .
= 8.5 cm to produce some polarized light some _ asymmetry _ is required .
in particular , if the radiation is due to the synchrotron process , the magnetic field can not be completely tangled , but must have some degree of order .
prior of the observational discovery , there have been theoretical studies predicting a larger degree of polarization ( @xmath60 : gruzinov & waxman 1999 ) due to causally disconnected region of maximally ordered magnetic field .
since the number @xmath61 of these regions is limited , the resulting polarization is of the order of @xmath62 , where 60% is the value reached in each single subregion .
this model requires a perfectly ordered magnetic field , on a scale that increases in time at almost the speed of light .
we ( ghisellini & lazzati 1999 ) have instead considered an alternative scenario , in which the required asymmetry is due to field compression and light aberration .
following laing ( 1980 ) , assume to compress a region embedded in an completely tangled magnetic field . after compression , the region becomes a slab : the field is squeezed " in one direction , and appears again completely random for face on observes ( with respect to the direction of compression ) , but highly ordered to edge on observers .
photons emitted in the plane of the slab can then be highly polarized .
if the slab moves with lorentz factor @xmath63 , those photons emitted in the slab plane ( perpendicularly to the direction of motion ) are aberrated in the observer frame , and make an angle @xmath64 with respect to the slab velocity .
observer looking at the moving slab at this angle will detect a large degree of optical polarization . consider a fireball collimated into a cone of semi aperture angle @xmath65 .
let @xmath66 be the angle between the cone axis and the line of sight . as long as @xmath67 , the observer receive radiation from a circle
entirely contained in the jet .
there is no asymmetry , and no polarization .
but when @xmath68 ( see fig .
2 ) , some position angles are not canceled , and some polarization survives .
the degree of polarization as a function of time is shown in fig .
3 : note the two maxima ( a third could be present if one considers side expansion of the jet which we neglected , see sari 1999 ) .
the position angles of the two maxima are orthogonal between each other .
3 also shows the light curve of the total flux , to illustrate the tight link that this model predicts between the existence of polarization and the gradual steepening of the light curve . should these ideas be confirmed
, we would have a very powerful tool to know the degree of collimation of the fireball , and hence the true total emitted power .
our knowledge of gamma
ray bursts had a quantum leap in the last two years .
but we do not know yet what is their main cause ( i.e. their progenitors , see the review by vietri 1999 , this volume ) , if it is the same for short and long bursts , what is the main radiation process of the burst , why their spectrum peaks at a well defined energy .
but the information in hand lead us to expect that they could be perfect cosmological probes , doing , for redshifts greater than 5 , what quasars have done for lower redshifts .
therefore there is a double interest to pursue grb research : to understand their nature and to use them to test the distant universe . for these reasons
another quantum leap is necessary , and it is foreseen to come from the launch of a dedicated nasa midex satellite , called swift , able to detect more than 100 burst per year and to slew to them with x ray and optical uv telescopes _ in less than 50 seconds ! _ ( see the dedicated web page at _ http://swift.gsfc.nasa.gov/_ ) . with these capabilities
, we will be able to have a large sample of accurate locations , and then , after the optical follow
up from the ground , we will have a large number of redshifts .
we will study the relatively soft x ray emission [ 0.110 kev ] and the optical emission while the burst is still on , and we will know the transition between the burst proper and the afterglow .
definitive approval of the mission is expected in the fall of 1999 , and launch is foreseen for the year 2003 .
the italian community is deeply involved and will provide the x
ray mirrors , developed by the brera observatory , and the malindi ground station , developed by asi for _ beppo_sax .
we think that this mission , besides being a cornerstone for the study of gamma ray bursts , will be a great opportunity for the italian community at large .
|
recent results of the hectic research activity about gamma ray bursts will be reviewed , with emphasis about the emission processes at the origin of the observed @xmath0rays .
the conventional synchrotron shock scenario is found to have problems , due to the very short cooling times of the emitting electrons , which implies a predicted spectrum , @xmath1 , much steeper than what is observed .
it is therefore compelling to look for alternative emission processes , such as quasi thermal comptonization , implying the presence of mildly or sub relativistic electrons , producing , through multiple compton scatterings , a spectrum @xmath2 ending with a wien peak where photons and electron energies balance .
the afterglow light , instead , can be indeed due to synchrotron radiation , and a confirmation of this is the recently detected optical linear polarization of the afterglow of grb 990510 .
some consequences of this discovery will be outlined . a quantum leap in our understanding of the physics of gamma
ray bursts is expected to come with swift , a nasa midex dedicated satellite .
# 1 # 2 _ mem . soc .
astron .
it . _ * # 1 * , # 2 # 1 # 2 _ the messenger _ * # 1 * , # 2 # 1 # 2 _ astron .
nach . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
lett . _ * # 1 * , l#2 # 1 # 2 _ astron .
astrophys .
rev .
_ * # 1 * , # 2 # 1 # 2 _ astron . astrophys .
suppl .
ser . _ * # 1 * , # 2 # 1 # 2 _ astron .
j. _ * # 1 * , # 2 # 1 # 2 _ ann .
rev .
astron .
astrophys .
_ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. lett .
_ * # 1 * , l#2 # 1 # 2 _ astrophys .
j. suppl .
_ * # 1 * , # 2 # 1 # 2 _ astrophys .
space sci .
_ * # 1 * , # 2 # 1 # 2 _ adv .
space res . _
* # 1 * , # 2 # 1 # 2 _ bull .
astron .
inst .
czechosl . _ * # 1 * , # 2 # 1 # 2 _ j. quant .
spectrosc .
radiat .
transfer _ * # 1 * , # 2 # 1 # 2 _ mon . not .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ mem .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ phys .
lett .
rev . _ * # 1 * , # 2 # 1 # 2 _ publ .
astron .
. japan _ * # 1 * , # 2 # 1 # 2 _ publ .
astr .
soc .
pacific _ * # 1 * , # 2 # 1 # 2 _ nature _ * # 1 * , # 2 epsf.sty
|
most stellar mass black hole binaries ( bhbs ) show common behaviour patterns centered around a few states of accretion .
the different accretion states are signified by different spectral energy distributions ( among other things ) having varying degree of contribution from the accretion disk and the non - thermal powerlaw components .
in addition , since _
chandra _ and xmm - newton , there has been increased interest in winds from the accretion disk , as a result of detections of blueshifted absorption lines of varying velocities and temperatures , seen in high resolution x - ray spectra . in order to get a consolidated picture of these systems ,
it is necessary to understand the relation between the accretion states of the bhbs and the properties of the accretion disk winds .
hetgs ( high energy transmission grating spectrometer , @xcite ) sensitive 0.5 - 7.5 kev energy range is rich in atomic transitions .
yet often absorption lines from only h- and he - like fe are detected ( e.g @xcite , @xcite for grs 1915 + 105 , @xcite for gx 339 - 4 , @xcite for h1743 - 322 and @xcite for igr j17091 - 3624 ) .
on the other hand in some rare cases a range of ions is seen , from o through fe ( e.g. @xcite for grs 1915 + 105 , @xcite for gro j165540 ) .
this suggests that the wind properties , e.g. temperature and density , may vary depending on the source and/or the accretion state . in this paper we restrict our discussion to low mass bhbs to ensure that wind signatures originate from the accretion disk rather than from the stellar companion ( as in the case of high mass x - ray binaries ) .
however , we note that accretion disk winds are detected in neutron star binary systems as well ( e.g @xcite , @xcite for cir x-1 , @xcite for gx 13 + 1 , @xcite for 1a0535 + 262 and @xcite for igr j17480 - 2446 ) and that some high mass x - ray binaries ( where the companion to the compact object is a high mass star ) exhibit stellar winds diagnosed through emission ( e.g. vela x-1 , @xcite ; lmc x-4 , @xcite ) or absorption ( e.g. cyg x-1 , @xcite ) .
x - ray studies of bhbs show that winds are not present in all states .
@xcite demonstrated that the equivalent width of the fe xxvi absorption line is anti - correlated with the fractional hard power law contribution to the spectrum in grs 1915 + 105 : the softer the state , the more prominent the absorption lines ; see also @xcite . recently
a confirmation of this finding came from the @xcite compilation of wind results ( by the aforementioned and other authors ) verifying the state dependent nature of accretion disk winds in general .
as an explanation for the observed changes in these winds , authors have commonly invoked differences in photoionizing flux ( e.g. * ? ? ? * for h1743 - 322 ) . however , since the properties of winds are also a result of their driving mechanisms @xcite , observed wind variations may also indicate changes in these driving mechanisms .
for example , a well known _
observation of gro j165540 @xcite , showing a rich absorption line spectrum with ions from oviii to nixxvi .
@xcite argue in favour of the magnetic driving mechanism for the wind . but a _ chandra _ observation from 3 weeks earlier , for the same object , shows only fe xxvi absorption @xcite .
the changes can not be explained by changes in the photoionizing flux alone , and @xcite suggest that variable thermal pressure and magnetic fields may both be important in driving long - term changes in this wind .
interestingly while changes in the photoionizing flux , in the driving mechanisms and in the wind density are frequently invoked to explain the observed wind behaviour , the thermodynamic stability of these outflows its relevance to the detectability of particular ions in bhbs have never been discussed in detail .
although quite commonly used for assessing winds in active galactic nuclei @xcite , thermodynamic stability arguments have not been used much to explain observed bhb wind behaviour .
( however , see @xcite for a specific use of the stability curves for determining accretion disk atmosphere properties for neutron star binaries . )
using thermal equilibrium curves , in this paper we test the importance of thermodynamic stability for the observable properties of winds in different accretion states of bhbs .
we are particularly interested in testing if the prevalence of the winds in the softer states ( as noted by * ? ? ?
* etc . ) can be explained using thermodynamic stability arguments .
we begin our considerations by adopting appropriate spectral energy distributions ( seds ) for different bhb states .
the sed of bhbs usually contains , to varying degrees , ( 1 ) a thermal component conventionally modeled with a multi - temperature blackbody originating from the inner accretion disk and often showing a characteristic temperature ( @xmath2 ) near 1 kev and ( 2 ) a non thermal power - law component with a photon spectrum @xmath3 @xcite .
the thermal state is dominated by the radiation from the inner accretion disk contributing more than 75% of the 2 - 20 kev flux ( e.g. dashed red line on top panel of ) .
in contrast the hard power law state ( a.k.a .
hard state ; solid black line in ) is dominated by a flat power law component ( @xmath4 ) that contributes more than 80% of the 2 - 20 kev flux . for any given bhb
, the accretion disk usually appears to be fainter and cooler in this hard power law state than it is in the thermal state .
the radiation from a thin accretion disk may be modeled as the sum of local blackbodies emitted at different radii .
a simple prescription for the radial distribution of the temperature is : @xmath5 @xcite where @xmath6 is the accretion rate of the central black hole of mass @xmath7 , @xmath8 is its eddington accretion rate and @xmath9 is the schwarzschild radius ( @xmath10 is the gravitational constant and @xmath11 is the velocity light ) . to describe the radiation from the accretion disk @xmath12
, we use the @xcite model ezdiskbb ( from xspec , @xcite ) .
the model ezdiskbb imposes the physical boundary condition that the viscous torque should be zero at the inner edge of the disk at radius @xmath13 .
the model is parametrised by @xmath14 and @xmath15 where is used to calculate @xmath16 ; @xmath17 is the distance of the source from us , @xmath18 is the angle that our line - of - sight makes to the normal to the disc plane ; @xmath19 is the hardening factor which accounts for the modification of the optically thick disk emission from a pure blackbody .
we use the established value @xmath20
@xcite we add a power - law component to the disk spectrum using @xmath21 \exp{^{-\frac{\nu}{\nu_{max } } } } \labequn{eqn : fullsed}\ ] ] to account for the full bhb sed , where @xmath22 is the spectral index of the powerlaw . in their review
, @xcite used the different seds of the 1996 - 97 outburst of the bhb gro j165540 ( hereafter groj1655 , * ? ? ? * ; * ? ? ? * ) to demonstrate the seds of a typical bhb in its different states .
following their prescription , we define three different fiducial seds for the three states of a @xmath23 black hole at a distance of 5 kpc from us .
( note that the the distance of the black hole from us does not have any consequence on the thermodynamic calculations presented in this paper for the absorbing gas which is very close to the black hole . to determine the photoionization state of the gas , it is sufficient to know the shape of the sed and the ionization parameter
. thus the exact value of the distance is not important . )
we also find in that a variation in @xmath24 does not strongly affect our overall conclusions . )
subsequent results presented in this paper will be based on the following sed definitions we adopt for the three typical low mass bhb states . 1 .
* thermal state * ( dashed red curve ) : to generate the disk spectrum , we choose @xmath25 resulting in @xmath26 ( @xmath2 ) = 0.48 ( 0.58 ) .
the powerlaw has @xmath27 and @xmath28 is chosen in such a way that the 2 - 20 kev disk flux contribution @xmath29 .
* intermediate state * ( dotted blue curve ) : the disk spectrum is generated with @xmath30 so that @xmath26 ( @xmath2 ) = 0.97 ( 1.17 ) .
for the powerlaw , @xmath31 and @xmath32 .
* hard powerlaw ( hereafter hpl ) state * ( solid black curve ) : with @xmath33 we generate a cooler disk with @xmath26 ( @xmath2 ) = 0.28 ( 0.33 ) .
the powerlaw is dominant in this state with @xmath34 and @xmath35 . for each of the seds defined above
, we use a high energy exponential cut - off ( ) to insert a break in the power law at @xmath36 kev ( ) .
see for a discussion of the effects of varying @xmath37 .
( in soft states @xmath38 can be less than 20 kev , and we accommodate this in our considerations of thermodynamic stability for soft states ; see for details . ) note that even though we base bhb state seds on gro j1655 ( following * ? ? ?
* ) , our results are intended to be generic to stellar mass black holes ( in so far as the 1996 - 97 outburst of gro j1655 was representative of a typical bhb outburst ) .
we use these seds as reasonable representations of the ionizing spectra for different bhb states , and we use similar representative parameters for the wind properties ( e.g. density , column density , and metallicity ) . in subsequent sections ,
we make quantitative tests of the influence of these choices , and find that our results are robust to changes in the sed and wind properties ( including @xmath39 in ; @xmath40 and @xmath41 and in ; @xmath37 in and wind density in .
in this section we use the aforementioned (; ) ionizing seds for the different bhb states to calculate the expected temperature , pressure ( ) of the outflowing gas , and the ionization fractions ( ) of the various ions in the gas .
this will allow us to assess the thermodynamic stability of winds as a means of explaining the presence / absence of observable ionic species in different bhb states . [
cols="^,^,^,^,^,^,^,^,^,^ " , ] + @xmath42 hi , n=2 @xmath43 photoionization from the n=2 level of neutral hydrogen + @xmath44 h iso . @xmath43 hydrogen iso - sequence + @xmath45 recomb .
@xmath43 recombination + @xmath46 uta @xmath43 unresolved transition array ( 2p-3d inner - shell absorption by iron m - shell ions ) thermodynamic stability of photoionized gas can be studied effectively using the thermal equilibrium curve of the temperature ( @xmath47 ) versus the pressure ( @xmath48 ) of the gas @xcite ; here @xmath49 is the ionization parameter . along the stability curve , heating and cooling are balanced everywhere , but gas in thermal equilibrium is only thermodynamically stable where the slope of the curve is positive .
the thermodynamic stability of the photoionized gas is determined by the balance of the major heating and cooling mechanisms along the curves ( see for an example list of such agents , varying with @xmath1 ) .
we expect that if the photoionized gas ( as diagnosed through observed absorption lines ) is detected , its ionization parameter is such that the gas is in thermal equilibrium .
hence the @xmath50 point of such gas should lie on the stable parts of the stability curve .
we model the photoionized absorber as an optically thin , plane parallel slab of solar metallicity @xcite gas with density @xmath51 and column density @xmath52 ( but see for a discussion on effects caused by density variation ) . using each of the three fiducial bhb seds as the ionizing continuum for this gas we generate three different stability curves ( ) using version c08.00 of cloudy@xcite . the stability curves span a wide range of @xmath1 values ( dotted black lines cutting across the curves ) , for most of which all the curves are stable .
however , we find that during the hpl state ( solid black ) , hot plasma with @xmath53 is thermodynamically unstable .
the ions associated with such gas are therefore unobservable .
in contrast , during the intermediate and thermal states , we find that the ionized gas is thermodynamically stable throughout @xmath54 , and thus a range of ions from low to high z ( atomic number ) elements can be detected in these states . all else being equal ( e.g. luminosity , density of the gas , and its distance from the central source ) , detections of ionized winds should be much more common in thermal and intermediate states .
our predictions of the thermodynamic properties of winds in bhb are consistent with the observations that winds are detected in softer , more thermal states and are rarely detected in spectrally hard states ( here represented by hpl , e.g. @xcite ) .
in we compare the major heating and cooling agents affecting the different bhb state stability curves in an effort to understand what causes the gas ionized by the hpl sed to become unstable at @xmath55 . as can be seen , at @xmath56 ,
compton heating is far more dominant for the hpl sed ionized gas ( 79% ) than for the thermal sed ionized gas .
this affects the temperature , which is higher , due to comptonization in the former case , and renders the hpl equilibrium curve unstable in the aforementioned @xmath1 range .
note that for the calculations in we have used a gas with constant density @xmath57 .
however , we tested the effect of varying the densities between @xmath58 in and find that the stability curves are _ insensitive _ to density variations for @xmath59 , leaving the above mentioned results for the range @xmath53 for the hpl state , unchanged .
the photoionization calculations performed in this paper are conducted in the static limit . in assessing the relevance of our conclusions for observed ionic differences between bhb states ,
we calculate the thermal cooling time scale @xmath60 ( defined as the time in which the gas looses half of the heat gained by it ; ) of the gas in the static limit . while calculating @xmath60 cloudy considers all the heating and cooling processes associated with static gas ( some of which are detailed in ) . the static gas assumption is reasonable if @xmath60 is less than the adiabatic cooling time scale @xmath61 ( @xmath62^{-1}$ ] , where @xmath63 , @xmath64 and @xmath65 are the acceleration , wind velocity , and radius respectively , @xmath66 being the boltzmann constant ) .
calculation of the adiabatic or dynamical time scale is beyond the scope of this paper ( but see @xcite for example discussions related to the dynamics of outflowing gas in agn and x - ray binaries ) .
we do not make any assumptions about the dynamics or the launching mechanisms of the gas .
the purpose of this paper is to study the thermodynamics of the photoionized gas , for which we have assumed that the static limit requirements are met . in this section
we further investigate the range @xmath53 in terms of ion fractions , to assess thermodynamic instability effects on ion observability .
in we plot the @xmath67 distribution of the ion fractions of the he - like and h - like ions of all the elements from zinc ( atomic no .
z = 30 ) to oxygen ( z = 8) .
vertical lines mark thermodynamically unstable range of @xmath67 based on our hpl ionizing sed . for a given ion
if a significant part of the hpl state ion fraction distribution falls within the unstable @xmath53 range , absorption lines of that ion would not be detected .
thus , during the hpl state a number of important he- and h - like species ( including fexxv ) are essentially ` shrouded ' by the thermodynamically unstable @xmath1 range .
purely from the thermodynamic point of view , these species are not expected to be visible during spectrally hard states .
it is notable , however , that the peak ion fraction for fexxvi ( and higher z h - like ions ) falls outside the unstable @xmath1 range . in other words , if there is a wind in a spectrally hard state with sufficiently high ionization ( @xmath68 ) , it could still be detected in fexxvi absorption .
the unstable @xmath53 range does not apply to the thermal / intermediate states , since there is no thermal instability .
as can be seen from , all ions should be observable from a thermodynamic stability point of view .
these results agree nicely with observational results on the long - term variability of winds .
as discussed above , winds are commonly detected through fexxv and fexxvi absorption during spectrally soft states , and rarely detected during harder states .
there are a small number of exceptions , however .
for example , in a 2005 _ chandra _ hetgs observation of groj1655 during a spectrally hard state , a single fexxvi absorption line was seen @xcite . @xcite explicitly note the absence of arxviii , sxv , and sixiv in that state . by our calculations ,
the absence of these ions is not surprising , given that the required @xmath1 for the gas producing these ions would have fallen within the thermodynamically unstable @xmath1 range .
the fiducial seds discussed in and shown in , although generic representations of the respective states , were generated using certain sed parameters . in this section we investigate the effects of varying sed parameters - @xmath69 ( ) , @xmath40 and @xmath41 ( ) and powerlaw break @xmath38 ( ) on thermodynamic stability results . we find that physically and observationally reasonable variations of these parameters do not alter the conclusions of our thermodynamic calculations ( ) . the same is true for variations in the wind density ( ) a plausible range of black hole masses lie between @xmath70 ( e.g. xte j1650500 @xcite ) and @xmath71 ( e.g see @xcite for grs 1915 - 105 , @xcite for m 33 x-7 and @xcite for cygnus x-1 ) . as discussed in ,
we choose a black hole of mass @xmath72 ( based on groj1655 ) to represent a generic bhb .
we further chose particular sets of @xmath73 and @xmath74 that correspond to values of @xmath26 and @xmath75 typically observed in outbursts of bhbs . given the relationship between temperature , disk radius , accretion rate , and black hole mass ( ) , how much are the seds affected by variations in @xmath7 ? in , we have compared the relationship between @xmath26 and @xmath75 for a @xmath23 black hole to that of a @xmath76 and a @xmath77 black hole .
the fiducial values for our three seds are marked as solid black stars .
the top panel of shows that for reasonable accretion rates and inner disk radii , our thermal / hpl seds could apply to a @xmath76 black hole ; the intermediate sed requires higher accretion rates ( @xmath78 ) , but not unheard of in bhbs .
the @xmath77 black hole can also match the fiducial values of @xmath26 with reasonable values of @xmath73 and @xmath74 , although the implied disk normalizations are a factor of @xmath79 larger .
thus we believe that our seds can be considered broadly representative of bhbs at a range of masses , accretion rates , and disk radii .
as discussed , we have followed the scheme of @xcite to pick reasonable combinations for accretion disk and power law components ( see ) .
since our stability curves are dependent on overall spectral shape , it is imperative to test if our results ( derived from the stability curves ) are robust to variations in @xmath40 and @xmath41 @xcite define the thermal state as having @xmath81 . keeping our @xmath41 constant at 2.5
, we calculated new stability curves for @xmath82 , and found no changes in the thermodynamic stability of the gas . since @xcite do not state a particular range for @xmath41 , we vary @xmath83 ( in steps of 0.2 ) keeping @xmath29 constant .
for @xmath84 the curve becomes unstable in the range @xmath85 .
however , a literature survey reveals no instances of @xmath86 in the thermal state .
hence we conclude that during the thermal state , winds should be thermodynamically stable for all values of @xmath1 . in addition
@xcite define the spectrally hard state ( which we call the hpl state ) as having @xmath87 and @xmath88 .
varying @xmath89 while keeping @xmath34 constant and then varying @xmath90 ( in steps of 0.2 ) while keeping @xmath91 constant we find that the curve retains the unstable @xmath1 range discussed in .
thus the qualitative results presented in hold for the entire range of the parameters of x - ray emission defined by @xcite for the bhb states . for all the seds discussed in the previous sections , the parameters for the exponential cut - off in the powerlaw ( ) were chosen such that the powerlaw had a break at @xmath92 .
observations of spectrally hard ( hpl ) states of bhbs show powerlaw breaks at @xmath92 , but for thermal states the break is usually observed at much lower energies @xmath93 .
it is believed that this has to do with the presence or absence of jets ( and associated particle acceleration processes ) in these states .
regardless of the origin of changes in the break energy , though , the effect on thermodynamic conditions is worth investigating , so we calculated new stability curves for @xmath94 , both for the thermal and hpl states . based on our analysis , the stability curves are insensitive to the variation in @xmath38 for @xmath95 ( k ) or @xmath96 . at higher temperatures ,
the stability curves become more smooth and the plasma actually becomes more stable as the break energy decreases from 100 kev to 20 kev .
this does not affect our conclusions for the thermal state , since the plasma was already stable at all @xmath1 . however , one consequence of variations in @xmath38 is that during the hpl state , the aforementioned range @xmath53 becomes thermodynamically stable ( see ) as @xmath38 is decreased to 20 kev .
the theoretical reason for this effect is easy to understand , where one recalls the discussion on heating and cooling agents in . according to , at high ionization parameters ( @xmath97 , the temperature of the gas ionized by hpl sed is determined by compton heating ( 79% contribution ) when @xmath98 .
however , if @xmath99 , then there is a significant decrease in the number of high energy photons thus reducing the extent of comptonization ( comptonization contributes only 49% to heating ) .
hence the temperature of the gas decreases resulting in a thermodynamically stable distribution of temperature and pressure .
although theoretically noteworthy , this is not a relevant result because the jet dominated hpl state is not likely to have such low break in the powerlaw .
thus , variation in the powerlaw break @xmath38 also does not affect the results in , unless there are reasons to consider that the hpl state continuum has a low powerlaw break at @xmath100 .
we vary the particle density of the absorbing gas between @xmath101 and generate stability curves ( ) for the hpl state ( left panel ) , the thermal state ( middle panel ) and the intermediate state ( right panel ) seds , to determine if the results presented in are susceptible to density effects . as can be seen from the figure ,
density variations do not have any effect on the stability curves for @xmath59 and the effect is small until @xmath102 . for lower @xmath67
the stability curves are seen to be thermodynamically unstable if the wind has density @xmath103 . because the vast majority of plasma in bhbs can be described by ionization parameters @xmath104 ( with the exception of gx 339 - 4 , @xmath105 ; @xcite ) , our conclusions for the thermodynamic stability ( or lack of it ) of winds are not affected by gas density variations .
even though we primarily see high @xmath106 gas in bhbs , we still consider density effects at low @xmath107 to improve theoretical understanding
. shows that the greatest difference in the gas temperature occurs at @xmath108 in comparing plasma with @xmath109 to one with @xmath110 . referring back to we find the detailed heating and cooling agents corresponding to the different stability curves .
specifically , comparing the results for @xmath111 in , the heating and cooling agents and their fractional contributions are similar for @xmath112 but differ from that of @xmath51 .
an explanation is that - line cooling is dominant for @xmath109 , whereas continuum processes like free - free cooling , recombination cooling and radiation through emission of iso - sequences of hydrogen and helium are dominant cooling mechanisms for @xmath51 resulting in higher temperatures in the later case .
we have constructed fiducial seds for the thermal , intermediate and hpl states of the bhbs , following the prescription of @xcite for a typical @xmath113 black hole , based on groj1655 . using these seds we generated thermodynamic stability curves for each state to arrive at the following conclusions : 1 . in the thermal and intermediate state
all phases of the wind is _ thermodynamically stable _ ( ) .
however , the ionization parameter range @xmath53 is _ thermodynamically unstable _ for winds in the hpl state .
we found that a large number of the he - like ions ( 21 @xmath114 z @xmath114 30 , ti through cu ) and h - like ions ( 15 @xmath114 z @xmath114 25 , s through mn ) have peak ion fractions in the unstable ionization parameter range for the hpl state , making the lines from these ions potentially unobservable ( ) .
our findings are well corroborated in the observational literature , since there appears to be a gradient in wind properties with bhb state . in accord with their thermodynamic stability ( and lack of it for
spectrally harder hpl states ) , winds are predominantly observed in intermediate and soft / thermal states and have weak or absent lines from fexxv and low - z elements in the hpl states @xcite . in the hpl state , while the ions with peak ion fractions in the unstable @xmath1 range may be unobservable , fexxvi may remain detectable at high ionization parameters ( @xmath115 ) in such states ( based on our calculations in and consistent with observations noted by @xcite ) .
3 . in general ,
a range of ionizations are thermodynamically stable ( except for @xmath53 for hpl state ) in all the states we have studied here . yet the observed absorbers are usually consistent with a single high ionization parameter .
this may indicate a genuine absence of plasma at lower ionization . on the other hand , for many other sources intervening cold absorption from the source or the ism may be the reason .
there have been suggestions for magnetically driven winds in e.g groj1655 @xcite .
the pressure term in the temperature - pressure distributions ( stability curves ) considered in this paper deals with only the thermodynamic gas pressure on the photoionised medium ( through @xmath48 ) .
magnetic pressure in not included in such calculations
. it would be interesting to incorporate the magnetic pressure term , which would change the stability conditions from what we have shown in this paper . given a reasonable description of the magnetic field in the photoionized medium , codes like cloudy are capable of such calculations . in our future publications
we shall formulate this phase space as a diagnostic tool for photoionized gas which are under the influence of strong magnetic fields .
we thank gary ferland for making cloudy publicly available and for providing helpful tips .
we acknowledge the generous support of the _ chandra _ theory grant tm3 - 14004x that is administered by the smithsonian astrophysical observatory .
chakravorty , s. , misra , r. , elvis , m. , kembhavi , a.k . , & ferland , g. , 2009 , mnras , 393 , 83 ferland , g. j. ; korista , k. t. ; verner , d. a. ; ferguson , j. w. ; kingdon , j. b. ; verner , e. m. 1998 , pasp , 110 , 761
|
we present a systematic theory - motivated study of the thermodynamic stability condition as an explanation for the observed accretion disk wind signatures in different states of low mass black hole binaries ( bhb ) .
the variability in observed ions is conventionally explained either by variations in the driving mechanisms or the changes in the ionizing flux or due to density effects , whilst thermodynamic stability considerations have been largely ignored .
it would appear that the observability of particular ions in different bhb states can be accounted for through simple thermodynamic considerations in the static limit .
our calculations predict that in the disk dominated soft thermal and intermediate states , the wind should be thermodynamically stable and hence observable . on the other hand , in the powerlaw dominated spectrally hard state the wind is found to be thermodynamically unstable for a certain range of @xmath0 . in the spectrally hard state , a large number of the he - like and h - like ions ( including e.g. fe xxv , ar xviii and s xv have peak ion fractions in the unstable ionization parameter ( @xmath1 ) range , making these ions undetectable .
our theoretical predictions have clear corroboration in the literature reporting differences in wind ion observability as the bhbs transition through the accretion states @xcite . while this effect may not be the only one responsible for the observed gradient in the wind properties as a function of the accretion state in bhbs , it is clear that its inclusion in the calculations is crucial to understanding the link between the environment of the compact object and its accretion processes .
physical data and processes - accretion , accretion discs , black hole physics , sources as a function of wavelength - x - rays : binaries , stars - ( stars : ) binaries : spectroscopic , stars : winds , outflows
|
for simplicity we consider a system that consists of one infinitely heavy quark ( or anti - quark ) and one light anti - quark ( quark ) .
the basic quantity for evaluating the energies of this heavy - light system on a lattice is the 2-point correlation function @xmath2 see fig .
[ fig_c2c3 ] a ) .
it is defined as @xmath3 where @xmath4 is the heavy quark propagator and @xmath5 the light anti - quark propagator .
@xmath6 is a linear combination of products of gauge links at time @xmath7 along paths @xmath8 and @xmath9 defines the spin structure of the operator . the @xmath10 means the average over the whole lattice .
the energies are then extracted by fitting the @xmath2 with a sum of exponentials , @xmath11 where the number of exponentials , @xmath12 , is 2 , 3 , or 4 and the time range @xmath13 .
@xmath14 is in practice a @xmath15 matrix for different fuzzings ( the indices are omitted in eq .
[ equ_c2 ] for clarity ) .
this fuzzing enables us to extract not only a more accurate value of @xmath16 , the energy of the ground state , but also an estimate of the first radially excited state , @xmath17 . after taking the continuum limit we can interpolate in the light quark mass and use charmed meson experimental results to interpolate in the heavy quark mass to get a @xmath0 meson .
the energy spectrum was measured using different lattices : q3 is a @xmath18 quenched lattice ( for more details see @xcite ) , df2 is a @xmath18 lattice and df3 , df4 are @xmath19 unquenched lattices .
df refers to dynamical fermions ( more details can be found in @xcite ) .
parameters of the lattices are given in table [ tbl_lattparam ] .
.lattice parameters . here
q means quenched and df means unquenched ( dynamical fermions ) . [
cols="^,^,^,^",options="header " , ] [ fig_c2c3 ] [ fig : energy ] from some theoretical considerations @xcite it is expected that , for higher angular momentum states , the multiplets should be inverted compared with the coulomb spectrum . if the potential is purely of the form @xmath20 the state l@xmath21 lies always higher than l@xmath22 .
here l@xmath21(@xmath22 ) means that the light quark spin couples to orbital angular momentum l giving the total @xmath23 .
( since the heavy quark is infinitely heavy its spin does not play a role here . )
however , in the case of qcd the confining potential is important for large distances and thus l@xmath22 should eventually lie higher than l@xmath21 for higher angular momentum states . experimentally this inversion is not seen for p - waves , and now the lattice measurements show that there is no inversion in the d - wave states either .
in fact , the d@xmath21 and d@xmath22 states seem to be nearly degenerate , _
i.e. _ the spin - orbit splitting is very small ( see fig . [
fig : energy ] ) . note that the spectrum in fig .
[ fig : energy ] shows an approximately linear rise in excitation energy with l.
[ fig_x11 ] [ fig_x12 ] for evaluating the radial distributions of the light quark a three - point correlation function is needed see fig .
[ fig_c2c3 ] b ) .
it is defined as @xmath24 we have now two light anti - quark propagators , @xmath25 and @xmath26 , and a probe @xmath27 at distance @xmath28 from the static quark .
we have used two probes : @xmath29 for the vector ( charge ) and @xmath30 for the scalar ( matter ) distribution .
the radial distributions , @xmath31 s , are then extracted by fitting the @xmath32 with @xmath33 see figs .
[ fig_x11 ] , [ fig_x12 ] .
the @xmath34 s and @xmath35 s are from the best fit to @xmath2 ( eq . [ equ_c2 ] ) . here
@xmath36 is the ground state distribution and @xmath37 , for example , is the overlap between the ground state and the first excited state .
[ fig_x11 ] compares the ground state vector ( charge ) distribution with the scalar ( matter ) distribution .
the matter distribution seems to drop off faster than the charge distribution .
however , at @xmath38 the charge and matter distributions are roughly equal . also note that the p - wave matter distribution changes its sign .
( hence the logarithmic scale in the s - wave plot but a linear scale in the p - wave plot in fig .
[ fig_x11 ] . ) in fig .
[ fig_x12 ] the charge and matter first excited state distributions are compared . as expected ,
a node is clearly seen in these excited state distributions .
the radial distributions shown here are for df2 , but the analogous calculations for df3 and df4 are in progress .
these distributions can be now treated as `` experimental data '' that needs understanding . .
here @xmath39 , @xmath40 and @xmath41 mev.,scaledwidth=37.0% ] [ fig_dirac2 ] [ fig_dirac ] the lattice measurements , _ i.e. _ the energy spectrum and the radial distributions , can be used to test potential models .
there are several advantages for model makers : first of all , since the heavy quark is infinitely heavy , we have essentially a one - body problem . secondly ,
on the lattice we know what we put in one heavy quark and one light anti - quark which makes the lattice system much more simple than the real world .
we can , for example , try to use the dirac equation to interpret the s- and p - wave distributions . if the solutions of the dirac equation are called @xmath42 ( the `` large '' component ) and @xmath43 ( the `` small '' component ) we get @xmath44 for the charge distribution and @xmath45 for the matter distribution .
these solutions are plotted in fig .
[ fig_dirac2 ] for a potential @xmath46 , where @xmath39 , @xmath40 and the mass that appears in the dirac equation is @xmath41 mev .
the values of these parameters are not tuned yet , because at this stage we only want to show that a reasonable set of parameters can give a qualitative description of the lattice data .
the qualitative agreement is surprisingly good see fig .
[ fig_dirac ] .
the dirac equation approach gives a natural explanation , for example , to the fact that the matter distribution drops off faster than the charge distribution .
also the change of sign in the p - wave matter distribution comes out automatically .
the key points are : * there should be several narrow @xmath0 states , in particular the @xmath47 that lies below the @xmath1 threshold . *
the spin - orbit splitting is very small , contrary to the expectation predicted by a linearly rising potential that is purely scalar . * the radial distributions of s and p@xmath22 states
can be qualitatively understood by using a dirac equation model .
|
the energies of different angular momentum states of a heavy - light meson were measured on a lattice in @xcite . we have now continued this study using several different lattices , quenched and unquenched , that have different physical lattice sizes , clover coefficients , hopping parameters and quark gluon couplings .
the heavy quark is taken to be infinitely heavy , whereas the light quark mass is approximately that of the strange quark . by interpolating in the heavy and light quark masses
we can thus compare the lattice results with the @xmath0 meson .
most interesting is the lowest p - wave @xmath0 state , since it is possible that it lies below the @xmath1 threshold and hence is very narrow .
unfortunately , there are no experimental results on p - wave @xmath0 mesons available at present .
in addition to the energy spectrum , we measured earlier also vector ( charge ) and scalar ( matter ) radial distributions of the light quark in the s - wave states of a heavy - light meson on a lattice @xcite .
now we are extending the study of radial distributions to p - wave states .
|
in this letter we consider diffusion in a multicomponent plasma of ions in dense stellar matter .
this diffusion can greatly affect the composition of matter particularly in cores of white dwarfs and envelopes of neutron stars .
it produces redistribution of ions ( heavier ions move to the star s center ) and extra energy release that reheats the star and affects its thermal evolution . for example , we can mention gravitational settling of @xmath2ne in carbon - oxygen ( @xmath3c@xmath4o ) cores of white dwarfs ( see , e.g. , refs.@xcite)which is thought to reheat old white dwarfs and helps explain observational data .
diffusion of ions affects also chemical evolution and nuclear burning in envelopes of neutron stars ( e.g. , refs .
transport properties are important also in dusty plasmas with their numerous applications ( e.g. , ref.@xcite ) .
diffusion equations are well studied in physical kinetics for the case when the ions constitute almost ideal plasma @xcite .
however , the ion plasma in white dwarfs and the neutron stars is typically strongly coupled by coulomb forces .
the diffusion coefficients of ions in strongly coupled coulomb plasma have been extensively studied in the literature , mainly by molecular dynamics simulations ( e.g. , refs.@xcite ) . here
we address a delicate problem of diffusion currents in strongly coupled coulomb plasma of ions .
consider a plasma which consists of electrons and a mixture of ion species @xmath5 , with atomic numbers @xmath6 and charge numbers @xmath7 .
let @xmath8 be the number density of ions @xmath9 .
the electron number density is @xmath10 ( due to charge neutrality ) .
it is convenient to introduce ( e.g. , ref .
@xcite ) the coulomb coupling parameter @xmath11 for ions @xmath9 , @xmath12 where @xmath13 is the temperature , @xmath14 is the boltzmann constant , @xmath15 is the electron - sphere radius , and @xmath16 is the ion - sphere radius ( for a sphere around a given ion , where the electron charge compensates the ion charge ) . therefore , @xmath11 is the ratio of a typical electrostatic energy of the ion to the thermal energy . if @xmath17 then the ions constitute an almost ideal boltzmann gas , while for @xmath18
they are strongly coupled by coulomb forces ( constitute either coulomb liquid or solid ) . one component ion plasma solidifies at @xmath19 .
we restrict ourselves to the gaseous or liquid ion plasma .
the diffusion currents in almost ideal plasma are well defined @xcite but the case of nonideal plasmas requires special attention .
we introduce these currents in the spirit of landau and lifshitz @xcite .
consider the matter which is slightly off thermodynamic equilibrium because of the presence of forces @xmath20 [ which act on all particles @xmath21 electrons ( @xmath22 ) and ions ( @xmath23 ) ] and number density gradients @xmath24 .
for simplicity consider isothermal matter ( no temperature gradients @xmath25 , and hence no deviations from thermal equilibrium ) .
the forces @xmath20 and gradients @xmath24 induce ( weak ) gradients of chemical potentials @xmath26 of particles @xmath21 .
let us introduce @xmath27 where we set @xmath28 ( @xmath29 and @xmath30 being electric charge and mass of particles @xmath21 , respectively ) .
the force @xmath20 is produced by gravitational acceleration @xmath31 ( that can be treated as a constant in the local approximation ) and the electric field @xmath32 due to weak plasma polarization in the gravitational field .
note that @xmath33 where @xmath34 is the mass density of the matter , and @xmath35 ( as prescribed by thermodynamics @xcite ) , @xmath36 being the ( total ) pressure .
the electric field drops off the sum because of electric neutrality but it is most important for driving different particle species ( e.g. , refs .
@xcite ) . if particles @xmath21 are in mechanical equilibrium , then @xmath37=0 .
this condition is exactly the same as the condition of chemical equilibrium used in ref .
if the plasma is in hydrostatic equilibrium as a whole , then @xmath38 and @xmath39 .
hydrostatic equilibration ( @xmath39 ) in neutron stars and white dwarfs is established over milliseconds
minutes @xcite but diffusive motion of ions can last over gigayears ( gyrs ) ( see , e.g. , ref .
this diffusion is studied by standard methods of physical kinetics @xcite assuming @xmath40 . in the diffusion problem
, a deviation of particles @xmath21 from mechanical equilibrium in matter can be conveniently measured by the vector @xmath41 where @xmath42 is the partial mass density of particles @xmath21 .
clearly , @xmath43 .
let @xmath44 be the diffusive flux of particles @xmath21 ( @xmath45 being the diffusion velocity of particles @xmath21 @xcite ) .
phenomenological transport equations can be written as @xmath46 where @xmath47 [ @xmath48 s@xmath49 can be called a generalized diffusion coefficient of particles @xmath21 relative to @xmath50 , and @xmath51 is a normalization function to be chosen later .
the diffusion coefficients should respect the relation @xmath52 in a rarefied , almost ideal plasma , we have @xmath53 , where @xmath54 is the partial pressure of particles @xmath21
. then eq .
( [ e : ja ] ) reduces to the standard definition of diffusion coefficients in rarefied gases @xcite . for strongly interacting particles , partial pressures @xmath54 are ambiguous , while the definition ( [ e : ja ] ) , based on chemical potentials @xmath55 , is not
. while the ions are heavy and slow , the electrons are light and mobile .
if we are interested in transport properties of ions , we can describe the electrons by the approximation similar to the born - oppenheimer approximation in the theory of molecules @xcite . specifically
, we assume that the electron gas is always in the state of mechanical ( quasi)equilibrium adjusting itself almost instantly to the motion of multicomponent ion plasma .
since the electrons are light , we can set @xmath56 .
then from eq .
( [ e : d ] ) we have @xmath57 .
therefore , the electron quasiequilibrium implies @xmath58 it allows us to factorize electrons out in the problem of ion transport ( diffusive fluxes of ions are mostly determined by a nonequilibrium state of the ion subsystem @xcite ) . in this case
( [ e : ja ] ) and ( [ e : sumja ] ) retain their form but indices @xmath21 and @xmath50 label only ion species ( @xmath5 ) .
note that eq .
( [ e : ja ] ) is strictly valid for nonrelativistic particles , whereas the electrons in dense matter can be relativistic .
however , the factorization works well even for relativistic electrons as long as they can be treated as massless . in the presence of two ion species ( @xmath9=1 ,
2 ) we have @xmath59 , @xmath60 , and @xmath61
. then kinetic phenomena can be characterized by one diffusion coefficient @xmath62 , @xmath63 here we have chosen @xmath64 ( @xmath65 being the total number density of the ions ) . then @xmath62 corresponds to the standard definition of the diffusion coefficient @xcite for two - component plasma of ions ( as follows from the equations presented below ) .
let us simplify eq .
( [ e : j1 ] ) . from eq .
( [ e : d ] ) we have @xmath66 with @xmath67 or 2 . because @xmath68 , we obtain the expression for @xmath32 : @xmath69 substituting it into ( [ e : d1 ] ) and setting @xmath70 ( @xmath71 being the atomic mass unit ) , we have @xmath72 .
\label{e : d1a}\end{aligned}\ ] ] quite generally , the chemical potential of ions @xmath9 is @xmath73 , where @xmath74 and @xmath75 label the ideal gas and coulomb contributions , respectively ( see , e.g. , ref.@xcite )
. then @xmath76 , with @xmath77 \nonumber \\ & = & \frac{k_bt}{n_e}\,\left(z_2n_2 \bm{\nabla}n_1-z_1n_1 \bm{\nabla}n_2 \right ) , \label{e : db } \\
\bm{d}_c & = & \frac{n_1 n_2}{n_e } \,\left[z_2 \bm{\nabla}\mu^{(c)}_1-z_1 \bm{\nabla}\mu^{(c)}_2\right ] .
\label{e : dc}\end{aligned}\ ] ] in eq .
( [ e : db ] ) we have used the well - known relation @xmath78 . combined with ( [ e : j1 ] ) , these equations give us the expression for @xmath79 .
it contains three terms labeled by subscripts @xmath80 , @xmath81 and @xmath82 .
the terms @xmath80 and @xmath81 are well known while the term @xmath82 seems new .
( @xmath80 ) .
assume that the matter is in hydrostatic equilibrium as a whole .
then in eq .
( [ e : da ] ) we have @xmath83 , so that @xmath84 describes gravitational sedimentation of the ions 2 ( provided their effective `` molecular weight '' @xmath85 is larger than that , @xmath86 , for ions 1 ) .
( @xmath81 ) .
the term @xmath87 is especially simple in the limit of @xmath88 .
then @xmath89 and @xmath90 which corresponds to ordinary diffusion of ions 2 .
generally , @xmath87 describes diffusive motion of the ions if their number densities are out of equilibrium .
( @xmath82 ) the term @xmath91 is most important in the regime of strong ion coupling and can be accurately described in the ion - sphere approximation combined with the linear mixing rule ( e.g. , ref .
@xcite and references therein ) : @xmath92 then @xmath93 the structure of @xmath91 is similar to that of @xmath84 : it describes specific ( `` coulomb '' ) sedimentation of ions 2 ( provided @xmath94 ) due to coulomb coupling in the gravitational field .
its remarkable feature is that it operates even for ions with @xmath95 .
such ions are commonly thought to have the same `` molecular weights . ''
then @xmath96 ( as long as we neglect small mass defects of ions 1 and 2 ) and one commonly assumes that such ions are not separated .
we see that it is not true . by way of illustration
let us rewrite eq .
( [ e : j1 ] ) under simplifying assumptions which are usually satisfied in the cores of white dwarfs and outer envelopes of neutron stars .
assume that the pressure is provided by strongly degenerate electrons , @xmath97 [ which allows us to express @xmath98 in eq .
( [ e : term3 ] ) through @xmath99 and the hydrostatic equilibrium is established ( @xmath100 ) .
then we obtain the diffusion flux in the standard form @xmath101 where @xmath102 are the velocities of gravitational settling of ions 2 due to `` molecular weight '' difference and coulomb separation , respectively ; @xmath103 .
note that , when the matter is in hydrostatic equilibrium , the gravitational settling of ions 2 is accompanied by `` lifting '' of ions 1 ( with @xmath104 ) . such diffusive motion of ions initiates collisional production of the specific entropy ( @xmath105 ) and the associated thermal energy release at a rate @xmath106 [ erg @xmath107 s@xmath49 ( e.g. , refs .
@xcite ) @xmath108 which is easily computed .
) of coulomb settling in isothermal degenerate cores ( @xmath109 ) of white dwarfs of different mass ( 0.2 , 0.6 , and 1.2@xmath110 , @xmath110 being the sun s mass ) and internal temperature ( 0.5 , 1 , and 5 @xmath111 k ) .
we show settling of @xmath3c in mixture with @xmath112he in @xmath113 star .
in other cases it is the settling of @xmath4o mixed with @xmath3c . in all cases
we assume that @xmath114.,scaledwidth=55.0% ] c ions mixed with @xmath115he ( @xmath116 ) in the outer envelope of a neutron star ( with gravity @xmath117 cm s@xmath118 ) versus depth @xmath119 ( measured from the stellar surface ) for two effective surface temperatures @xmath120 and @xmath121 k. while @xmath119 varies from 1 to 100 m , the density increases from @xmath122 to @xmath123 g @xmath107.,scaledwidth=55.0% ] although the diffusion flux ( [ e : i2 ] ) has standard form , it contains a new gravitational settling term ( [ e : uc ] ) due to coulomb separation .
this separation has been predicted by chang , bildsten and arras @xcite who considered equilibrium distributions of ion mixtures including the coulomb interaction term .
thus we extend their work to nonequilibrium mixtures and show that the coulomb separation is pronounced in the diffusion flux ( [ e : i2 ] ) and drives gravitational settling of ions .
the most pronounced effect occurs at temperatures at which the ions constitute strongly coupled coulomb liquid . at lower temperatures the ions
solidify and diffuse much more slowly @xcite . at higher @xmath13 coulomb coupling
is weak and less efficient ( although generally available )
. the coulomb sedimentation should be especially important for the mixtures of ions with the same @xmath0 ( for instance , mixtures of @xmath115he , @xmath3c , and @xmath4o ions ) . the traditional gravitational sedimentation ( [ e : ua ] ) in such mixtures
is greatly suppressed ( can occur only due to mass defects of atomic nuclei @xcite ) .
the coulomb settling in these mixtures ( [ e : uc ] ) is typically much stronger than ( [ e : ua ] ) .
the ions with larger @xmath1 should move to deeper layers .
the effect is stronger for a larger difference of @xmath1 in the mixture . because the effect is primarily driven by gravitational forces , it should be most pronounced in compact stars ( white dwarfs , and especially in neutron stars ) with strongest gravity .
first of all we mean @xmath115he@xmath3c cores of low mass white dwarfs and @xmath3c@xmath4o cores of more massive white dwarfs , and evolution of similar mixtures in the envelopes of neutron stars .
the velocity of sedimentation is given by eq .
( [ e : uc ] ) using appropriate diffusion coefficients ( e.g. , refs.@xcite , and references therein ) .
figure [ fig : wd ] presents the velocity @xmath124 in the @xmath3c@xmath4o cores of medium mass and massive white dwarfs and in the @xmath115he@xmath3c cores of low mass white dwarfs .
the adopted temperate range @xmath125 k is appropriate to rather old white dwarfs ( e.g. , ref .
the settling velocities are higher for massive white dwarfs ( with larger @xmath126 ) .
the velocity profile throughout the core has a maximum at the core boundary @xmath127 , where the gravitational acceleration @xmath128 is the largest .
the velocity @xmath129 as @xmath130 because @xmath131 at the star s center .
the maximum velocity in the massive ( @xmath132 ) white dwarf reaches @xmath133 km gyr@xmath134 , meaning that the @xmath3c@xmath4o separation in the outer core can occur in a few gyrs .
the velocity of coulomb separation of @xmath3c and @xmath4o ions is typically lower than the settling velocity of @xmath2ne ions in the @xmath3c@xmath4o core @xcite , but the fraction of @xmath2ne ions is much smaller than the fractions of @xmath3c and @xmath4o . using eqs .
( [ e : uc ] ) and ( [ e : q ] ) we have estimated the thermal energy generation rate @xmath135 which accompanies this separation and found it insufficiently high to noticeably reheat old white dwarfs .
the profile @xmath135 has maximum in the outer part of the white dwarf core .
note that our estimates neglect the direct diffusion term [ the first term in eq .
( [ e : i2 ] ) ] which can enhance @xmath135
. the coulomb separation of @xmath115he , @xmath3c , and @xmath4o ions can be important in isolated and accreting white dwarfs .
it affects chemical composition and , therefore , microphysics of white dwarf core ( heat capacity , thermal conductivity , neutrino emission , nuclear reaction rates ) as well as chemical , thermal , and nuclear evolution of white dwarfs .
redistribution of ions due to coulomb separation can affect also vibration properties of stars ( asteroseismology ) . coulomb separation of ions with equal @xmath0 in neutron star envelopes is much stronger than in white dwarfs .
figure [ fig : ns ] plots the sedimentation velocity @xmath124 of @xmath3c ions mixed with @xmath112he in the outer neutron star envelope versus depth @xmath119 ( measured from the surface ) for two effective surface temperatures , @xmath136=1 and 2 mk .
the temperature profile @xmath137 within the envelope has been determined by solving the heat transport equation for a conserved heat flux emergent from stellar interior ( see , e.g. , ref.@xcite ) .
the envelope is nonisothermal and the temperature gradient can affect diffusion which we ignore for simplicity .
therefore , the presented curves should be treated as illustrative .
for the densities of @xmath138 g @xmath107 ( a few to a few tens of meters under the surface ) the sedimentation velocity can reach a few meters per year .
the separation can affect nuclear evolution of the matter in the outer layers of accreting neutron stars . it will change the thermal conductivity of this matter , influence the relation between the surface and inner temperatures of neutron stars and affect cooling of isolated and accreting neutron stars ( see , e.g. , refs .
@xcite , and references therein ) .
we are grateful to a. i. chugunov and a. y. potekhin for useful discussions .
d. g. y. acknowledges support from rfbr ( grants no .
11 - 02 - 00253-a and no .
13 - 02 - 12017-ofi - m ) , rf presidential program nsh 4035.2012.2 , and ministry of education and science of russian federation ( agreement no . 8409 , 2012 ) .
|
we analyze diffusion equations in strongly coupled coulomb mixtures of ions in dense stellar matter . strong coupling of ions in the presence of gravitational forces and electric fields ( induced by plasma polarization in the presence of gravity ) produces a specific diffusion current which can separate ions with the same @xmath0 ( mass to charge number ) ratios but different @xmath1 .
this coulomb separation of ions can be important for the evolution of white dwarfs and neutron stars .
|
the schechter luminosity function @xcite has been the standard expression for representing the space density of galaxies over the past 25 years .
it has strengths and weaknesses .
the strengths are : its connection to the fundamental theory of the growth of initial mass perturbations @xcite , its overall simplicity ( with three free parameters : the characteristic luminosity @xmath2 ; the normalisation @xmath3 ; and the faint - end parameter @xmath4 ) , and its simple analytical connection to the luminosity - density ( and ultimately via a mass - to - light ratio to the galaxian matter - densities ) .
its weaknesses are : the correlation of the three parameters , the critical dependence of all three parameters upon the `` turn - over '' region , and its inability to reflect deviations of the space - density from a simple power - law at faint luminosities . over the past two decades
numerous attempts have been made to constrain the three defining schechter function parameters in both field and cluster environments producing a range of conclusions from a ubiquitous luminosity function to strong environmental dependencies .
here we showcase selected recent results from a variety of methodologies , to address this specific question as to whether the galaxy luminosity function is ubiquitous or not .
so let us start with the field or global luminosity function . fig .
1 shows a compendium of eight recently derived @xmath5 or @xmath6-band luminosity functions transformed to a common bandpass , @xmath7 ( see liske et al 2003 ) . deriving mean values and errors from these data
yield : @xmath8 mag , @xmath9 , @xmath10mpc@xmath11 , providing a crude idea of the systematic uncertainties ( recall the values are strongly correlated ) .
clearly the precise global survey one adopts will be critical for any field / cluster comparison .
so which survey to adopt ?
this really depends upon the likely causes of these variations .
most likely @xmath4 is under - constrained due to insufficient statistics at the faint - end , compounded by the luminosity - surface brightness relation ( driver 1999 ) .
this can result in lower luminosity galaxies being preferentially missed in shallower surveys ( see cross & driver 2002 ) .
@xmath12 is presumably due to calibration errors and @xmath3 to cosmic variance . under these latter assumptions
one can adopt the two largest surveys ( the 2dfgrs ; norberg et al 2002 and the sdss2 ; blanton et al 2003 ) as the most credible . however
although these now agree at the bright end they show discrepancy at the faint - end ( see fig .
.,width=453,height=283 ] one extremely important point , particularly with respect to comparisons with cluster lfs , is the very limited absolute magnitude range over which the global lf is known . to highlight this fig .
2 shows the _ observed _ distribution of the largest galaxy redshift survey to date ( the 2dfgrs ; colless et al 2001 ) .
the 2dfgrs contains 221,233 galaxies of which only 9,398 galaxies are fainter than @xmath13 ( see fig .
2 ) . with an incompleteness of 8% the resulting luminosity distribution can only be considered absolutely secure , i.e. robust to any spectroscopic completeness bias , over a fairly restricted magnitude range @xmath14 .
this inability to pin down the faint - end slope via a direct approach severely restricts the range of comparison between the field and cluster environments . that the two most recent and largest redshift surveys show significant discrepancy in @xmath4 is obviously a cause for concern .
one entirely orthogonal way to constrain @xmath4 which is _ robust _ to most selection effects , has recently been demonstrated by liske et al ( 2003 ) .
they use the _ curvature _ of the bright millennium galaxy catalogue precision number - counts alone to constrain @xmath3 and @xmath4 by adopting a prior for the mean characteristic luminosity from the latest redshift surveys ( @xmath15 ) .
the justification is that the redshift surveys optimally sample the @xmath2 point and , with this well defined , the curvature of the counts now depend entirely upon @xmath4 ( assuming the cosmology , k - corrections and evolutionary corrections are known ) .
3 shows the 1- 2- and 3-@xmath16 confidence contours in the @xmath17-plane .
the mgc results agree remarkably well with the 2dfgrs ( see fig . 1 ) and we therefore adopt the 2dfgrs / mgc parameters as our global yardstick ( see table 1 ) .
we now move to the luminosity functions of the near and dear .
3 shows a compendium of the most recent local group , local sphere , virgo , fornax and coma luminosity functions ( see fig . 3 caption for references ) . where possible the most recent and largest datasets have been selected to reflect a measure of the entire cluster lf .
in particular we note that in all cases the group / cluster membership is based on either direct distance indicators ( local group and local sphere ) , eyeball assessment ( virgo and fornax ) or spectroscopy ( coma ) as opposed to blind background subtraction . also shown ( fig .
3 near right and far right ) are the 1-,2- and 3- @xmath16 error - ellipses derived from a simple @xmath18-minimisation of the schechter function .
3 ( near right ) are those derived from fitting the entire available magnitude range and fig .
3 ( far right ) are those derived from fitting only over a luminosity range equivalent to the known global lf range ( see figs 1 & 2 ) . from fig .
3 ( far right ) we must conclude , contrary to some studies , that all environments are consistent with each other and the global luminosity function .
however the analysis is limited by the insufficient number of members in the few clusters studied compounded by the limited range over which the global luminosity function is known .
perhaps more compelling is the apparent consensus between the overall luminosity functions from the sparse local group to the dense coma cluster .
and @xmath19 error ellipses for a schechter function fit to either the full luminosity range ( @xmath20 various , middle ) or the range comparable to the field range ( @xmath21 , right ) , all fits are to the solid data points . local group data :
pritchet & van den bergh 1999 ( solid ) ; mateo 1998 ( open ) , local sphere data : jerjen , binggeli , freeman 2000 ( solid ) ; karachtenseva et al 2002 ( open ) , virgo data : trentham & hodgkin 2001 ( solid ) ; fornax data : ferguson 1989 + deady et al 2003 ( solid ) ; coma data : mobasher et al 2003 ( solid ) ; beijersbergen et al 2002 ( open triangles ) , trentham 1998 ( open circles ) and andreon & culliandre 2002 ( open squares )
. the solid line shows , in each case , the optimal schechter function fit over the full range of data .
the data point and errorbars on the right most panels shows the location of the adopted global luminosity function.,width=453,height=604 ] the coma cluster is worth some brief further comment as there have been a number of recent studies with intriguing and even conflicting results . on the largest scale beijersbergen et al ( 2002 ) sample 5.2 sq degrees using background subtraction to determine the overall lf finding an anomalously low @xmath12 ( @xmath22 ) when compared to the 1 sq degree spectroscopic study of mobasher et al ( 2003 ) .
conversely mobasher et al find a significantly flatter faint - end slope than beijersbergen et al .
more restricted surveys such as trentham ( 1998 ) and andreon & culliandre ( 2002 ) which focus on the core find evidence for an upturn and/or dip inconsistent with the larger area studies ( see fig .
however beijersbergen et al show that when they confine themselves to the trentham region they also recover the identical dip and upturn as seen by trentham .
a plausible explanation from a close reading of the relevant papers is the following .
beijersbergen et al may have over estimated their field counts at the bright - end , ( see their table 4 column 6 which is flatter than the expected @xmath23 relation at bright mags ) , this will have minimal impact upon their core lf where the cluster contrast is higher hence agreement with trentham but results in a significant bright - end over - subtraction of the full cluster area hence disagreement with mobasher et al .
meanwhile mobasher et al may suffer from some b - band incompleteness in their r - band selected spectroscopic sample .
finally the trentham and andreon & culliandre lfs are dominated by the core population and the additional structures ( lfs dips , upturns ) must therefore be core phenomena resulting from strong evolutionary processes in the core region . clearly the definitive study of coma remains to be done and represents an excellent opportunity for a space - based survey where high - resolution imaging can be used to determine accurate cluster membership .
for clusters beyond coma the majority of studies have relied on the background subtraction of the field galaxy population to recover the cluster luminosity function to comparably faint absolute magnitudes ( e.g. , driver et al 1994 ) . as the discussion on coma above suggests the method of background subtraction , while efficient , is susceptible to a variety of errors . in driver
et al ( 1998 ) we discuss the criterion for successful background subtraction in substantial detail .
in essence the process is only reliable for very rich clusters where the contrast of the cluster population is high .
valotto , moore & lambas ( 2001 ) more recently argue for a bias towards steeper faint - end slopes because of preferential alignments between clusters and filaments . while this argument has some merit locally it does not hold for clusters at intermediate redshifts @xmath24 .
the reason for this is that structures are correlated only over scales of order 100mpc minimising the impact upon clusters at appreciable distances .
however valotto et al do highlight the potential impact of cosmic variance in general and some analysis have neglected to include this significant error component[multiblock footnote omitted ] . compounding this issue is also the ad hoc coverage of the various studies .
in particular the majority of deep studies which have recovered lfs to faint absolute magnitudes sample only the core region identifying features such as dips and upturns similar to that seen in coma .
overall the conclusions fall into two opposing camps , those that find significant variations between clusters ( for example garilli et al 1999 ; goto et al 2002 and qv ) and those that do not ( for paolillo et al 2001 and qv ) .
the derived composite lfs for these three surveys are shown in table 1 .
so are the variations real or an artifact due to factors such as areal coverage , cosmic variance and/or surface brightness incompleteness between the cluster and reference fields ? until direct membership is ascertained via either spectroscopy , photometric redshifts or high - resolution space - based imaging
it is unlikely that results from background subtraction can be taken as conclusive either way .
from the above it is clear that , for the moment , the only reliable way to fully assess the environmental impact on the lf is from within a single large spectroscopically confirmed and self - consistent catalogue .
the 2dfgrs offers such an opportunity ( see de propris et al 2003 ) .
the 2dfgrs is a magnitude limited redshift survey to an approximate limit of @xmath25 with 221233 redshifts and 92% completeness .
the survey has median redshift of @xmath26 and contains 60 well sampled nearby rich clusters ( see de propris et al .
full details of the analysis are given in de propris et al ( 2003 ) however fig .
4 shows the end result comparing the derived field ( solid line ) and composite cluster ( open circles ) lfs along with the error ellipses . note that the field and composite cluster lfs are based upon @xmath27 and 4,186 spectroscopically confirmed galaxies respectively .
the formal schechter function fits are also shown in table 1 .
these suggest a marginally brighter @xmath12 and marginally steeper @xmath4 . perhaps more important though de propris et al go on to subdivide their sample of 60 clusters according to ; rich versus poor , high versus low velocity dispersion , bautz - morgan class i , i - ii , ii versus ii - iii , iii and inner versus outer . in all cases
the composite lfs remain consistent with the original composite lf , although it is noted that the composite _ inner _ lf is poorly fit by a schechter function indicating an upturn comparable to that seen in the core coma studies .
statistically though the only significant variation comes when the field and cluster lfs are subdivided according to type . in these case
it is seen that the early - type galaxies have both a slightly brighter and steeper lf while the late - type galaxies remain unchanged in shape but reduced in terms of relative normalisations , see fig .
at face value we must conclude that the galaxy luminosity function appears ubiquitous across all environments studied with only marginal evidence for a slightly brighter @xmath12 and a slightly steeper @xmath4 in clusters ( see also the article in these proceedings by christlein ) . only in the very core regions of rich clusters or when the population is sub - divided according to spectral type are significant variations seen .
we postulate that these results lead to three likely conclusions : \(1 ) cluster infall is an ongoing process with a substantial fraction of the cluster population infalling in recent times .
this explains the universality of the lf from the field to rich clusters and constrains any epoch of major merging to have occurred prior to the epoch of cluster formation .
\(2 ) star - formation is effectively and efficiently inhibited in the infalling population however the halo merger rate is low .
this explains the variation with spectral type but not in the overall lf .
hence while galaxies may have shifted their spectral classes their broad - band output remains mostly unchanged .
\(3 ) dramatic evolutionary processes ( merger , harassment etc ) resulting in the construction of the cd / d galaxies and destruction of the dwarf population is confined to the core region .
this explains the dips and upturns seen in coma , the 2dfgrs core sample and numerous deep background subtracted core studies .
this is a data rich time for this field with the development of wide field mosaics , the advanced camera for surveys and the upcoming large - format infrared detectors .
we look forward to the power of these technologies being brought to bear on this unfolding story .
we acknowledge invaluable access to the two - degree field galaxy redshift survey database and note that the phenomenal success of the 2dfgrs is largely based on the hard work and dedication of the staff of the anglo - australian observatory .
we thank the organisers of the jenam2002 galaxy evolution workshop for a very enjoyable meeting .
|
we review selected measurements of the galaxy luminosity function including the global field , the local group , the local sphere , nearby clusters ( virgo , coma and fornax ) and clusters in general .
we conclude that the overall cluster luminosity function is consistent with the global luminosity function over the magnitude range in common ( @xmath0 ) .
we find that only in the core regions of clusters ( @xmath1 kpc ) does the overall form of the luminosity function show significant variation .
however when the luminosity function is subdivided by spectral type some further variations are seen .
we argue that these results imply : substantial late infall , efficient star - formation suppression , and the confinement of mass - changing evolutionary processes to the core regions only .
|
the spin @xmath0 kondo model describes a localized magnetic moment interacting with an electron gas , via an antiferromagnetic exchange coupling.@xcite despite a long history , and even an exact solution , it has not yet surrendered all its secrets .
it is well established that the ground state is a spin singlet in which the impurity spin is quenched by the electron gas , and the spatial region where the electron gas is correlated with the impurity is referred to as the kondo screening cloud.@xcite even for an isotropic system , its precise spatial profile is not known analytically , except asymptotically,@xcite and it is only in the past few years that it has been calculated numerically.@xcite despite wide - ranging proposals,@xcite it has eluded direct measurement , partly because of the difficulty in measuring spin correlations . in the isotropic case ,
the screening cloud is characterized by the ground state correlation function @xmath1 where @xmath2 is the impurity spin operator , and @xmath3 is the electron spin density at @xmath4 .
when the kondo temperature is much lower than the fermi energy , the screening cloud can be decomposed into a forward scattering contribution @xmath5 and a backscattering contribution @xmath6 , so that @xmath7 , where the two functions @xmath5 and @xmath6 vary slowly on the scale of the fermi wavelength @xmath8 . in the scaling regime and for spin - isotropic kondo exchange ,
the profile of the screening cloud displays a universal line shape,@xcite which is dependent on the value of kondo coupling @xmath9 only through an emergent length @xmath10 that is inversely proportional to the kondo temperature.@xcite to be specific , if @xmath11 and @xmath12 , with @xmath13 , are correlation functions corresponding to different values @xmath9 and @xmath14 of the kondo coupling , and @xmath10 and @xmath15 are the corresponding kondo lengths , then @xmath16 for all @xmath17 . recently , we have realized that it may be possible to measure the longitudinal forward scattering ( @xmath18 ) component of the screening cloud ( a precise definition is given below ) in a chain of tunnel - coupled superconducting islands.@xcite it turns out that the hamiltonian describing the charge and phase degrees of freedom in this system is equivalent to the _ spin - anisotropic _ kondo model in one dimension,@xcite described by two coupling constants : the @xmath19-components of the impurity and electron spins couple with a strength @xmath20 , while the components perpendicular to the @xmath19-axis couple with a different strength @xmath21 .
a detailed study of the screening cloud in the anisotropic kondo model has to our knowledge not been performed before , and is therefore timely . in the anisotropic case ,
there are four correlation functions of interest , namely @xmath22 , that measures the correlation between @xmath19-components of the impurity and electron spins ( both in the forward @xmath23 and backward @xmath24 scattering channels ) , and @xmath25 , that measures correlations between components perpendicular to the @xmath19 axis ( in each channel ) .
clearly , an anisotropic kondo interaction , characterized by two coupling constants @xmath20 and @xmath21 , will affect the universal scaling picture non - trivially .
the universal scaling of the isotropic model has been confirmed using the numerical renormalization group ( nrg ) , a popular method to study the kondo model.@xcite however , there is still room to improve the accuracy of existing results .
for instance , it is predicted analytically @xcite that backscattering dominates forward scattering inside the screening cloud .
this produces oscillations from ferromagnetic to antiferromagnetic correlations between the impurity and the electron gas , on the scale of half the fermi wavelength .
the first numerical calculation of the universal scaling functions were reported in ref . .
however , the predominance of the backscattering component over the forward scattering component inside the cloud was not resolved .
this was probably due to the fact that results were obtained at relatively large kondo temperatures , so that there was not a sufficient separation of scales between the size @xmath10 of the cloud , and the short distance ultraviolet cut - off scale . in a more recent work,@xcite the nrg method of ref .
was refined , and the alternation of ferro- and antiferromagnetic correlations inside the cloud clearly be seen .
however , results were only presented for distances up to @xmath26 fermi wavelengths from the impurity . at these scales ,
the correlation functions show non - universal modulations . in this article , we perform numerical calculations that are sufficiently accurate to investigate the universal scaling behavior in regimes ranging from isotropic to strongly anisotropic kondo couplings . at the same time
, we can clearly resolve the predominace of the backscattering component over the forward scattering component inside the cloud .
in precise terms , we investigate the the following questions : let @xmath27 and @xmath28 , with @xmath29 and @xmath13 , be correlation functions at distinct values @xmath30 and @xmath31 .
under which conditions are there constants @xmath32 and @xmath33 such that @xmath34 , for all @xmath17 , and what are the line shapes of these universal scaling curves ?
our general numerical findings show that scaling is typically obeyed for fixed values of @xmath20 . in other words , correlators at the same @xmath20 but different @xmath21 can be scaled onto each other .
this statement holds as long as @xmath35 ( in dimensionless units of the inverse density of states ) , while for transverse couplings of order one , the kondo temperature becomes comparable to the fermi energy , and universality is lost . for @xmath36 , we find that the line shapes of the cloud correlation functions acquire a @xmath20-dependence that can not be scaled away . however , as @xmath20 approaches @xmath21 , the @xmath20 dependence rapidly becomes very weak .
one thus find a sizable region of parameter space , located around the strict spin - isotropic line @xmath37 , in which the universal curves have little discernible @xmath20 dependence . in this region , correlators calculated at @xmath38 _ and _ @xmath39 can to a very good approximation
be scaled onto the unique universal curve of the isotropic model .
these conclusions are consistent with poor man s scaling arguments @xcite at small @xmath21 , as we will explain in detail .
our work also elaborates on a different but related topic , that was raised very recently by lukyanov _
_ in ref . , where the overlap between two kondo wave functions with different kondo couplings was computed analytically using integrability techniques , and numerically with the density matrix renormalization group .
we complement these interesting results , by providing a simpler ( approximate ) analytical expression for the kondo overlaps in the spin - anisotropic limit .
we also show that our numerical method reproduces the full analytical result , as it should .
regarding methodology , our approach combines both analytics and numerically exact calculations . in the case of strong spin - anisotropy , @xmath40
, we provide asymptotically exact formulas , that were not discussed in previous literature , for the fermionic kondo cloud . for a general choice of parameters
, we employ a recent approach using a coherent state expansion of the wave function , @xcite that has not yet been applied to the computation of purely fermionic observables that pertain to the kondo model .
this method relies on a variational ansatz , formulated in the language of the spin - boson model , which describes a two - level system coupled to a multi - mode ohmic bosonic bath.@xcite this spin - boson model is known to be equivalent to the anisotropic kondo model,@xcite with bosonization providing an exact mapping between the two.@xcite we exploit this mapping to apply the coherent - state expansion to the kondo model .
in addition , the flexible structure of the ansatz allows one to add progressively more contributions to it , so that the method converges rapidly to the exact ground state with arbitrary accuracy . historically , the systematic coherent state expansion proposed by bera _
et al . _
@xcite builds on seminal works of emery and luther,@xcite and of silbey and harris,@xcite in the language of the spin - boson model , and on independent works by anderson,@xcite and bergmann and zhang@xcite in the language of the kondo model .
the approach of emery and luther , silbey and harris , and anderson corresponds to the lowest order approximation , involving a single coherent state , and provides _ quantitatively _ accurate results only in the limit @xmath41 .
bergmann and zhang took a step in the direction of a two coherent - state ansatz by adding an extra term to the lowest order approximation , but constrained its form sub - optimally compared to the full solution with two coherent states .
we stress that this type of approach differs from the well - known method pioneered by yosida,@xcite that takes as starting point a state in which a single electron binds into a singlet with the impurity , while the rest of the fermi sea is unaffected .
the wave function of the bound electron is chosen to maximize the binding energy .
the relationship between the yosida approximation and the single - coherent state ansatz is reminiscent of the relationship in superconductivity between the single cooper pair on top of a fermi sea on the one hand , and the full bcs wave function on the other .
yosida s approach gives qualitatively correct results , but even when the effect of additional particle - hole excitations are included,@xcite it can not yield arbitrarily accurate results for the correlations inside the kondo screening cloud . in this article
, we also contribute an important innovation to the coherent - state expansion methodology . in previous implementations ,
the bosonic bath was limited to a finite number of modes , because the number of variational parameters in the ansatz scaled linearly with the number of bath modes . here
, we first partially solve the variational problem analytically , so that the number of remaining variational parameters that have to be determined numerically is independent of the number of bath oscillators .
this allows us to take the thermodynamic limit before we optimize the energy numerically , and probe arbitrarily large distances . in the kondo language
, this means that we work directly with an infinite conductor . with nrg ,
probing large length scales comes at the cost of losing information about shorter lengths scales . in order to maintain good spatial resolution ,
one has to introduce a fictitious impurity at the position where correlators are evaluated .
each position considered then requires the solution of a given two - channel kondo impurity problem , so that a high - resolution computation of the kondo cloud with nrg is quite expensive numerically .
our method remarkably deals with all length scales on an equal footing , so that the kondo cloud can be calculated in a single step , once the many - body ground state is known .
the coherent state expansion thus nicely complements the existing tools for studying the kondo model .
the rest of the article is structured as follows . in sec .
[ secham ] we define the anisotropic kondo hamiltonian .
we also state the equivalent spin - boson model , and the relation between the parameters of the two models .
the precise mapping between them is reviewed in appendix [ appmap ] . in sec .
[ sco ] we define the spatial correlation functions , that contain information on the entanglement cloud around the spin impurity , and that we will study . we first consider the correlations in the fermionic language of the kondo model and then in the bosonic language of the spin - boson model .
the bosonic observables are derived from the fermionic ones in appendix [ appcloud ] .
[ secmpa ] introduces the systematic coherent state expansion .
the innovation we mentioned in the previous paragraph , that reduces the number of variational parameters to solve for numerically , is presented in sec .
[ secreduce ] . in sec .
[ secen ] we take the thermodynamic limit for the energy , and express it in terms of the remaining variational parameters .
the correlation functions that we study are expressed in terms of the remaining variational parameters in sec .
[ cloudcalc ] .
next , we find the optimal values for the variational parameters by minimizing the energy , and we present our results in sec .
[ secresults ] .
we first confirm the accuracy of the method and then study in great detail the various correlation functions describing the kondo cloud , as well as the kondo wavefunction overlaps .
finally , in sec .
[ secdisc ] we summarize our main findings and , where applicable , compare them to results in the literature .
in this section we write down the two related models that we study , and make explicit the deep connection between them .
the first hamiltonian is the anisotropic spin-@xmath0 kondo model in one dimension , that describes a spin-1/2 impurity coupled to a fermi sea .
the fermi sea does not necessarily represent electrons confined to one dimension .
if the electron gas is higher dimensional , but the impurity is a point - like scatterer , it only interacts with electronic @xmath42-waves .
such @xmath42-waves in the electron gas can then be described by a one - dimensional kondo hamiltonian , where the spatial coordinate is the radial distance from the impurity .
the second hamiltonian is the ohmic spin - boson model , that appears in many area of physics , from superconducting nano - circuits to biological systems . below an ultraviolet energy scale set by the fermi energy ,
the two models are equivalent.@xcite the full mapping is reviewed in appendix [ appmap ] .
the anisotropic spin-@xmath0 kondo model in one dimension reads @xmath43 , where : @xmath44,\nonumber\\ h_\perp&=&\frac{j_\perp^b}{2}\left[\tilde \psi^\dagger_\uparrow(0)\tilde \psi_\downarrow(0)\sigma^- + \tilde \psi^\dagger_\downarrow(0)\tilde \psi_\uparrow(0)\sigma^+\right]\label{eq1}.\end{aligned}\ ] ] here @xmath45 annihilates an electron of wave number @xmath46 and spin direction @xmath47 on a ring of length @xmath48 , and @xmath49 .
we denote these operators with tildes , because we will denote the useful slow modes ( defined below ) without tildes .
we assume @xmath50 , so that there are both left and right - movers at the fermi energy .
the superscript @xmath51 indicates the bare values of the coupling constants @xmath52 and @xmath53 .
integrating out high energy degrees of freedom towards the fermi surface , we end up with a linear dispersion relation , and an effective low energy hamiltonian : @xmath54,\nonumber\\ h_\perp&=&j_\perp\left[\psi^\dagger_\uparrow(0)\psi_\downarrow(0)\sigma^- + \psi^\dagger_\downarrow(0 ) \psi_\uparrow(0)\sigma^+\right],\label{h2}\end{aligned}\ ] ] in units where the fermi velocity @xmath55 .
we have defined slow modes @xmath56 and a similar relation between @xmath57 and @xmath58 . here
@xmath59 is the ultraviolet cut - off energy .
operators without over - bars are associated with even - parity single particle wave functions , so that @xmath60 creates an electron in the state @xmath61 $ ] . in @xmath62 dimensions ,
an equivalent role is played by @xmath42-waves . for negative @xmath4
, @xmath63 creates an electron in an even - parity state consisting of two wave packets centered around @xmath64 and propagating in opposite directions towards the impurity . for positive @xmath4 on the other hand
, it creates an electron in a state consisting of two wave packets propagating away from the impurity .
because @xmath63 and @xmath65 create electrons localized to the same physical regions in space , we refer to ( [ h2 ] ) as the unfolded representation .
operators with over - bars are associated with odd - parity single particle wave functions , so that @xmath66 creates an electron in the state @xmath67 $ ] .
since the even and odd modes decouple , and only the even modes couple to the impurity , we focus on electrons in even orbitals and drop the @xmath68 terms in @xmath69 . in the process of linearizing the spectrum ( integrating out fast modes ) , the bare couplings @xmath52 and @xmath53 are renormalized to new values @xmath20 and @xmath21 , that depend on the ultraviolet scale @xmath59 .
note however that su(2 ) symmetry , if present , is preserved under renormalization .
thus , if @xmath70 ( isotropic limit ) , then @xmath71 . in this work
, we will make extensive use of the well - known fact , reviewed in appendix [ appmap ] , that the kondo hamiltonian ( [ h2 ] ) can be mapped onto an ohmic spin - boson model : @xmath72 here @xmath73 are bosonic operators such that @xmath74=0 $ ] and @xmath75=\delta_{q , q'}$ ] , while @xmath76 and @xmath77 are pauli matrices .
these pseudo - spin operators are not the physical angular momentum operators of the kondo impurity , but are related to them .
the parameters of the spin - boson model are related to those of the kondo model by @xmath78
having defined the system we investigate , we now write down the observables that we will study .
we are interested here in ground state correlation functions between the impurity spin and the electron spin density at a distance @xmath4 from the impurity , which we refer to collectively as the kondo screening cloud ( or cloud for short).@xcite we give equivalent expressions for the cloud correlators as ground state expectation values in both the fermionic and the bosonic pictures . in appendix [ appcloud ]
the bosonic expressions are derived starting from the fermionic expressions . in the fermionic picture , the cloud correlation functions is defined as : @xmath79 where @xmath80 is the @xmath81 component of the impurity spin operator , and @xmath82_{\sigma\sigma ' } \tilde \psi_{\sigma'}(x)\ ] ] is the @xmath83 component of the electron spin density at @xmath4 .
the subscript k indicates that the expectation value is with respect to the fermionic kondo ground state .
we refer to @xmath84 as the longitudinal and @xmath85 as the transverse cloud . both consist of a component @xmath86 that varies slowly on the scale of the fermi wavelength , and a component that oscillates with wave vector @xmath87 , and has an amplitude @xmath88 .
the former is the result of scattering events that change the electron momentum by an amount that is small compared to @xmath89 , while the latter results from scattering between the fermi points at @xmath90 .
explicitly , one has : @xmath91 x^\parallel_{2k_f}(x ) .
\label{cloud1}\end{aligned}\ ] ] at @xmath97 , the behavior of these expressions depend strongly on ultraviolet physics that our model does not attempt to represent accurately .
the regime of physical significance is @xmath92 where the behavior of the cloud is insensitive to ultraviolet details of the model .
none the less , we delay taking the @xmath98 limit until the end of the calculation , reasoning that it is of some interest to see how a particular cut - off scheme regularizes ultraviolet singularities .
we presented expressions for a one - dimensional cloud .
however , the generalization to a higher dimensional electron gas is trivial , assuming a point impurity that therefore only scatters @xmath42-waves.@xcite in this case , one interprets the coordinate @xmath4 as a radial distance and divides the one - dimensional correlator by the area of a spherical shell of radius @xmath4 .
one also reverses the sign of the @xmath87 components , because the radial part of the higher dimensional problem is defined on the half - line , with a @xmath99 phase shift between left- and right movers . to make further progress in the computation of the cloud observables
, we need an accurate approximation for the ground state of the spin - boson model .
that is the topic of the next section .
we will use here a systematic coherent - state decomposition @xcite of the many - body ground state of the ohmic spin - boson model . physically , coherent states appear as natural degrees of freedom , because , in the absence of spin - tunneling @xmath100 , the two spin configurations are associated with a displacement of the bath modes from @xmath101 to @xmath102 .
indeed , at @xmath103 , there are two degenerate ground states @xmath104 where @xmath105\left|0\right>,\ ] ] @xmath106 is the bosonic vacuum , and @xmath107 are the spin eigenstates of @xmath77 .
the single coherent - state ansatz ( usually dubbed the silbey - harris state ) includes the effect of tunneling by promoting @xmath108 to a variational parameter and taking as trial state the linear combination @xmath109 .
this results readily in the simple expression @xmath110 with the self - consistency condition @xmath111 .
this trial wavefunction provides an excellent approximation to the true ground state provided the shifted equilibrium positions @xmath112 of most oscillators are not too far apart , _
i.e. _ if @xmath113 is small compared to unity . for larger @xmath113 ,
the overlap @xmath114 , with @xmath108 determined variationally , becomes exponentially small , and this strongly underestimates the tunneling energy @xmath115 .
the coherent - state expansion addresses this issue by extending the silbey - harris form to a more general linear superposition of coherent states : @xmath116 with the set of displacement @xmath117 parametrizing a family of silbey - harris states : @xmath118 \left|0\right>. \label{family}\ ] ] here @xmath119 is the maximal number of allowed coherent states in the decomposition ( [ superpose ] ) , and sets the level of approximation . the displacements @xmath120 and coefficients @xmath121
are determined by minimizing @xmath122 where @xmath123 is a lagrange multiplier that is used to enforce normalization .
these parameters are generically found to be real in the ground state .
the coherent - state expansion already dramatically improves the estimate for the tunneling energy @xmath124 for @xmath125 , by allowing for cross - terms @xmath126 , in which the displacements @xmath117 and @xmath127 are anti - correlated at low @xmath128 .
such cross - terms are therefore not exponentially small in @xmath113 , and allow a sizable energy gain compared to the silbey - harris approximation .
it is important to note that an arbitrary state of the spin - boson hamiltonian can be written as an infinite but _ discrete _ sum of the form ( [ superpose ] ) , at least if the bosonic bath contains a finite number of modes .
this follows from a theorem , proved by cahill,@xcite that for a single bosonic mode , countable sets @xmath129 of ( real ) coherent states exist , that form a complete basis . for a finite number of bosonic modes , a general state can therefore be approximated to any required accuracy , by making @xmath119 sufficiently large .
it is in this sense that the coherent state expansion is numerically exact , provided in practice that good convergence to the true many - body ground state occurs .
previous investigations @xcite , as well as the extensive comparisons made in sec .
[ secresults ] , demonstrate indeed that the expansion ( [ superpose ] ) rapidly approaches the exact ground state of the spin - boson model as @xmath119 is increased to moderate values , also for an infinite bath .
in previous implementations of the coherent expansion ( [ superpose]-[family ] ) , the bosonic bath was restricted to a large but finite number @xmath130 of modes , and each of the @xmath131 displacements @xmath117 was treated as an independent variational parameter . in this work ,
we want to calculate the kondo screening cloud , which requires a high spatial resolution from short to possibly exponentially large distances , so that a huge number of bath modes needs to be included .
it is therefore desirable to have an implementation of the method in which the number of variational parameters does not depend on the number of included modes .
such an implementation would also allow for computations for a continuous spectral density ( thermodynamic limit ) , which numerical techniques like nrg are not able to perform . in this section ,
we derive such an implementation , which is the the main technical innovation of our work .
we show that the algebraic structure of the variational equations constrains the functional dependence of the displacements @xmath120 with respect to momentum @xmath128 to such an extent that there are in fact only @xmath132 variational parameters , independently of the number of modes @xmath130 of the bosonic bath . to stress that the argument does not require a linear bath spectrum , we replace the term @xmath133 in the spin - boson hamiltonian , with the more general kinetic energy expression @xmath134 for the remainder of this section . in subsequent sections
, we will specialize again to the linear bath spectrum . exploiting the coherent state structure of the ansatz
, we can express the total energy @xmath135 in terms of @xmath120 and @xmath121 as : @xmath136-\frac{\delta}{2}\left < f^{(m)}\right| \left .- f^{(n)}\right>\right\},\label{ef}\ ] ] where @xmath137.\label{ol}\ ] ] considering the minimization condition @xmath138 with the use of ( [ ef ] ) and ( [ ol ] ) , this leads to @xmath139 where the entries of the @xmath140 matrices @xmath141 and @xmath142 , and the @xmath119-dimensional column vector @xmath143 are : @xmath144,\nonumber\\ & q_{lm}=c_m\frac{\delta}{2}\left < f^{(l)}\right|\left .
-f^{(m)}\right>,\nonumber\\ & w_l=\frac{1}{2}\sum_m u_{lm}.\label{mat}\end{aligned}\ ] ] it is crucial to note that @xmath141 , @xmath142 and @xmath143 in ( [ sys ] ) do not depend on the value of the momentum @xmath128 labelling the displacement @xmath117 that we consider within the variational equation .
these matrices and vectors however depend non - linearly on the complete set of displacements , through summations over a dummy momentum index as in eq .
( [ ol ] ) . from ( [ sys ] )
, it readily follows that @xmath117 is given by : @xmath145_{mn}w_n.\ ] ] since @xmath146 is linear in @xmath147 , @xmath148_{mn}w_n=\frac{n^{(m)}(\omega_q)}{d(\omega_q)},\ ] ] where @xmath149 is a polynomial of order @xmath150 in @xmath19 and @xmath151 is a polynomial of order @xmath119 in @xmath19 . note that the denominator @xmath152 is common to all sets of displacements @xmath153 for @xmath154 . using the expression ( [ mat ] ) for @xmath143 in terms of @xmath141 , we see that for large @xmath19 , @xmath155 independent of @xmath156 .
thus we arrive at the form : @xmath157 where @xmath158 , and in general @xmath159 is real .
the poles @xmath160 that are not real come in complex conjugate pairs .
the optimal state @xmath161 can be obtained by numerically minimizing the energy with respect to the @xmath162 unknown coefficients @xmath159 , the @xmath119 poles @xmath160 , and the @xmath150 weights @xmath121 ( accounting for wavefunction normalization ) , so that there are @xmath132 parameters to be found in total , independently of the number of bosonic modes in the problem . since the dimension of the search space does not depend on the number of bath modes ,
the main obstacle in considering a bath with a very large number of modes has been removed .
the next question is whether the energy , which involves sums over all bath modes , can be calculated efficiently for a large number of modes .
only single sums over the momentum have to be computed in the energy functional , and in the worst case , the total numerical cost is linear in the number of modes .
this is a major improvement with respect to a brute force diagonalization of the model , where scaling of the hilbert space dimension is exponential with the number of degrees of freedom . in order to reach the continuum limit ,
the discrete momentum sums are replaced by integrals , which fortunately can be performed analytically rather than numerically , as we now demonstrate .
in this section we specialize again to the case of a linear bath spectrum @xmath163 , and analytically perform the momentum integrals involved in the calculating the energy for given set of variational parameters @xmath159 , @xmath164 and @xmath121 .
we do so for the smooth ultraviolet cut - off @xmath165 that the bosonization of the kondo model naturally introduces .
although we do not work out the details here , we note that the momentum integrals can also be performed analytically in the case of a sharp cut - off @xmath166 . in the @xmath167 ( thermodynamic )
limit , momentum sums are replaced by integrals according to @xmath168 with @xmath120 of the form ( [ e7 ] ) , and @xmath169 of the form ( [ mappar ] ) dictated by the mapping from the kondo model , the energy @xmath170 can be written as a functional of the variational parameters as : @xmath171 which involves the laplace transforms @xmath172 of the rational functions @xmath173 ^ 2,\nonumber\\ j_{mn}^0(k)&=&\left[kh_m(k)-1\right]\left[kh_n(k)-1\right].\end{aligned}\ ] ] note that we have omitted here a constant term @xmath174 that does not depend on the variational parameters .
the functions @xmath175 have @xmath119 second order poles of the type @xmath176 . for large @xmath46 ,
the behavior is @xmath177 the laplace transforms can be performed analytically , using the following method . in general
, we consider integrals of the form @xmath178 where the function @xmath179 is a ratio of polynomials , and all poles are of second order , i.e. @xmath180 this excludes the possibility of poles on the positive real line , which yields @xmath181 . furthermore
, the numerator @xmath182 is a polynomial of at most order @xmath183 . ) .[f1 ] ] it would be useful if we could replace the integration contour in ( [ e1 ] ) by a closed contour such as the one in figure [ f1 ] .
we can do so , provided we multiply the integrand by a function @xmath184 with a branch cut along the positive real line such that @xmath185 .
furthermore , @xmath186 must be such that the contribution to the integral that comes from closing the contour at large @xmath46 is negligible . a function that exactly meets these requirements is @xmath187 , where @xmath188 is the incomplete gamma function of order zero.@xcite it is defined as @xmath189 where the integration path does not intersect the negative real line and excludes the origin .
its main relevant properties are @xmath190 for @xmath191 , and @xmath192 for large @xmath193 .
furthermore @xmath194 is analytical except in the neighborhood of the positive real line .
thus we find @xmath195 in the first line , @xmath196 is the contour depicted in figure [ f1 ] .
returning to the integrals @xmath197 , one then has @xmath198 setting @xmath199 and noting that @xmath200,\label{res1}\ ] ] evaluation of the residues yield @xmath201 where @xmath202.\end{aligned}\ ] ] to relate the @xmath203 parameters to the coherent state displacements explicitly , we define @xmath204 and @xmath205\nonumber\\ & = & \frac{\sum_{l=1}^{m-1}l\mu_{ml}\omega_n^{~l-1}}{\prod_{l=1\not = n}^m(\omega_n-\omega_l ) } -h_{mn}\sum_{l=1\not = n}\frac{1}{\omega_n-\omega_l}.\nonumber\\\label{defhd}\end{aligned}\ ] ] this gives @xmath206 substitution of ( [ ee5 ] ) into ( [ e4 ] ) , and ( [ e4 ] ) into ( [ en ] ) yields an explicit expression for the energy in terms of the variational parameters . as a consistency check
, we can work out the single coherent state theory ( @xmath207 ) to make sure that we recover the results of silbey and harris .
this approximation is accurate for @xmath208 , which , in the kondo language , corresponds to the strongly anisotropic @xmath41 regime . for @xmath207 , there is only one variational parameter , which is conventionally defined as @xmath209 .
this translates to @xmath210 and @xmath211 [ see eqs .
( [ defh ] ) and ( [ defhd ] ) ] .
the energy of the silbey - harris state then reads @xmath212,\ ] ] with @xmath213\right\},\ ] ] and we have used the identity ( [ res1 ] ) to relate @xmath214 and its derivative . minimizing @xmath215 with respect to @xmath216 , we recover the known self - consistency condition @xmath217 assuming than @xmath218 , and using the fact that for small @xmath19 , @xmath219 where @xmath220 is the euler - mascheroni constant , one can solve the self - consistency condition for @xmath216 to obtain @xmath221 up to a pre - factor , @xmath216 corresponds to the kondo temperature in the regime of small @xmath113 , where the single coherent state ansatz is accurate . in general , it is far more efficient to evaluate the energy analytically , as was done in this section , than to do the integrals in ( [ en ] ) numerically .
there are however small regions of the search space where the analytical evaluation of the energy is not stable .
these are regions in which two or more of the poles @xmath164 lie close to each other . in these regions , there are large cancellations between individual terms in the sum over residues in ( [ resexp ] ) , and this leads to large numerical errors when individual residues are calculated separately before they are summed .
we circumvent the problem as follows .
when the minimization algorithm searches a dangerous region of the search space , it does not try to evaluate the residues at the offending poles individually .
it rather takes the slow but safe option of numerically integrating around a loop that circles all closely spaced poles at a safe distance .
fortunately , one does not have to fall back on this contingency plan too often , as the problematic regions of the search space are small and do not seem to be particularly favored in the actual optimal solution .
in the previous section , we obtained an analytical expression for the energy in terms of @xmath132 variational parameters .
this result allows for a significant speed - up of the numerical minimization of the energy . in this section
, we apply the same analytical technique to evaluate the momentum integrals involved in the calculation of the kondo screening cloud , in terms of bosonic displacements . evaluating the four different cloud correlators ( [ bosoncloud ] ) for the @xmath119-coherent state wavefunction
can be done using straight - forward coherent - state algebra : @xmath222}\cos(\sqrt{\alpha}b(x)_{mn})c_mc_n \left < f_m\right|\left .- f_n\right>,\nonumber\\ x_{2k_f}^\perp(x)&=-\frac{a}{\pi(x^2+a^2)}\sum_{m , n=1}^me^{\sqrt{\alpha}[a(0)_{mn}-a(x)_{mn}]}c_mc_n \left < f_m\right|\left .-
f_n\right>,\nonumber\\ x^\parallel_0(x)&=-(1-\sqrt{\alpha})\frac{a}{\pi(x^2+a^2)}+\frac{\sqrt{\alpha}}{2\pi}\sum_{m , n=1}^m c(x)_{mn}c_mc_n\left < f_m\right|\left.f_n\right>,\nonumber\\ x^\parallel_{2k_f}(x)&=\frac{2x}{\pi(4x^2+a^2)}{\rm im}\left[\frac{a+ix}{a - ix}\sum_{m , n=1}^m e^{i\sqrt{\alpha}d(x)_{mn}}c_mc_n\left < f_m\right|\left.f_n\right>\right ] , \label{cloudeqs}\end{aligned}\ ] ] together with the normalization condition @xmath223 the matrices @xmath224 , @xmath51 , @xmath196 , and @xmath225 are defined as : @xmath226,\nonumber\\ b(x)_{mn}&=\int_0^\infty dq\ , \sin(qx)e^{-aq}[h_m(q)-h_n(q)],\nonumber\\ c(x)_{mn}&=\int_0^\infty dq\ , \cos(qx)e^{-aq}[qh_m(q)+qh_n(q)-2],\nonumber\\ d(x)_{mn}&=\int_0^\infty dq\ , \sin(qx)e^{-aq}[h_m(q)+h_n(q)].\label{cint}\end{aligned}\ ] ] note that @xmath51 , @xmath196 and @xmath225 remain finite if the limit @xmath227 followed by @xmath228 is taken , while @xmath224 diverges logarithmically .
the method we used to evaluate @xmath229 analytically can be extended to evaluate the above integrals as well .
the detail of the calculation can be found in appendix [ appci ] .
the resulting expressions are : @xmath230 where @xmath231 is defined in ( [ efl ] ) . for future reference
, we now consider the large ( compared to the kondo length ) @xmath4 asymptotic behavior of the expressions ( [ cloudeqs ] ) for the screening cloud . at @xmath232 , @xmath233
the residue theorem can be used to derive the identities @xmath234 with the aid of the above equations , the cloud correlators ( [ cloudeqs ] ) are found to decay like @xmath235 at large @xmath4 .
explicitly , one finds for @xmath236 : @xmath237 where the lengths @xmath238 and @xmath239 are given by @xmath240c_mc_n \left <
f_m\right|\left.f_n\right>.\label{ea1}\end{aligned}\ ] ] the above expression establishes a very direct connection between the @xmath241 behavior of the coherent state displacements and the large @xmath4 behavior of the longitudinal cloud .
the small contribution @xmath242 to @xmath243 is a feature of the smooth ultraviolet cut - off @xmath244 , and will be ignored in what follows .
note that the @xmath245 and @xmath87 components of the cloud become equal at large @xmath4 .
furthermore , there are two emergent length scales @xmath238 and @xmath239 in the problem , which are in general not equal , unless spin - isotropy is restored . in the single coherent - state approximation , which is accurate for small @xmath113 ,
the cloud correlators can be explicitly computed : @xmath246},\nonumber\\ & x_0^\parallel(x)=-(1-\sqrt{\alpha})\frac{a}{\pi(x^2+a^2)}-\frac{\sqrt{\alpha}\delta_r}{\pi}{\rm re}\,f(-\delta_r(a - ix)),\nonumber\\ & x_{2k_f}^\parallel(x)=\frac{2x}{\pi(a^2 + 4x^2)}{\rm im}\left[\frac{a+ix}{x - ix } e^{2i\sqrt{\alpha}{\rm im}\,f(-\delta_r(a - ix))}\right].\label{shcloud}\end{aligned}\ ] ] where , as noted in sec .
[ secen ] , @xmath209 is the kondo energy scale .
the two correlation lengths @xmath238 and @xmath239 [ cf .
( [ ea1 ] ) ] are in this anisotropic limit : @xmath247 these simple analytical expressions for the kondo cloud , valid in the limit of strong spin - anisotropy , have not appeared in the literature before .
we will analyze them further in the next section , along with numerical results obtained for larger @xmath113 values from the systematic coherent state expansion .
in sec . [ secen ] and
[ cloudcalc ] , we have collected the tools to calculate the average energy and the components of the screening cloud , for an @xmath119 coherent - state wavefunction , in terms of only @xmath248 variational parameters . in order to find the variational parameters , the energy must be minimized , and for this purpose we use a standard simulated annealing algorithm.@xcite clearly , the quality of the approximation is limited by the maximum number of coherent states that can be handled with the available computational resources . using a single personal computer and simulated annealing minimization ,
we have found it possible to go up to @xmath249 coherent states , which is enough for our purposes here , although an improved algorithm combining global and local optimization @xcite can reach values as large as @xmath250 .
in fact , the required number of numerical operations is not a limitation _ per se _ here .
rather , the main difficulty is that the energy landscape in the space of variational parameters is very shallow and contains several low lying minima .
this section is divided into three extended sub - parts .
first , we benchmark the coherent state method , both against previous results , and also by studying the convergence properties of the coherent state expansion , establishing its domain of validity for the computation of the screening cloud .
then , we consider the kondo overlaps proposed in ref . , which we compute both analytically in the spin - anisotropic regime , and numerically for the nearly spin - isotropic case .
our results quantitatively agree with recent field theoretic calculations.@xcite we finally perform an extensive study of the kondo cloud correlators over a wide range of parameter values by a combination of analytical arguments and extensive coherent state simulations .
this allows us to uncover the precise universal features of the screening cloud of the anisotropic kondo model . in the minimum energy @xmath251 as the number of coherent states @xmath119
is increased , at three different points in parameter space .
circles correspond to a strongly spin - anisotropic point @xmath252 , triangles to a moderately spin - anisotropic point @xmath253 , and diamonds to a perfectly spin - isotropic point @xmath254 , associated to equal kondo exchange couplings @xmath255 . ] as a function of the number @xmath119 of coherent states , for the same three points in parameter space as the data in figure [ f2 ] .
[ fenvar ] ] in this subsection , we present strong evidence for the rapid convergence of the wavefunction ( [ superpose ] ) as the number @xmath119 of coherent states increases , using various observable quantities . in refs . and
, convergence was established only for a spin - boson model containing a large but finite number of modes , and we demonstrate here that fast convergence also occurs for a continuous bath spectrum .
we start by plotting in figure [ f2 ] the fractional improvement @xmath256 in the minimum energy @xmath251 obtained by adding an extra coherent state in the ansatz , as a function of @xmath119 .
note that , as in sec .
[ secen ] , we do not include the constant @xmath257 in the definition of @xmath251 .
this convention leads to a denominator in @xmath256 that is closer to zero , and therefore to a more stringent measure of convergence .
figure [ f2 ] shows results for three points in parameter space .
one of the points , @xmath258 , corresponds to a strongly anisotropic situation where convergence is very rapid .
another curve @xmath259 corresponds to a less anisotropic situation , where the convergence is slower .
the remaining point @xmath260 corresponds to the isotropic coupling @xmath255 , or a kondo temperature @xmath261 . for @xmath252 , the minimum energy changes by an amount comparable to the accuracy goal of the minimization module , by the time that @xmath119 reaches @xmath262 . for the other two points in parameter space ,
the marginal change in the minimum energy is a fraction of a percent at @xmath249 .
all the results that we present below are at least as converged as these last two cases . as a complementary test
, we can also verify that the multiple coherent state @xmath161 converges to an eigenstate of @xmath263 as the number of coherent states increases .
to do so , we calculate the energy uncertainty @xmath264 for the optimal coherent states ansatz @xmath161 , as a function of @xmath119 . for this purpose ,
an expression for the overlap @xmath265 is needed .
it turns out that the same integrals @xmath197 as in sect .
[ secen ] are involved . in terms of these integrals ,
the overlap is expressed analytically as : @xmath266 the rest of the calculation can then proceed with the technology developed in sect .
[ secen ] for evaluating the integrals @xmath267 .
note that derivatives @xmath268 , refer to partial derivatives with respect to the explicit @xmath269 dependence of the integrals , and not to total derivatives involving the implicit @xmath269-dependence of the variational parameters .
the energy uncertainty @xmath270 is showed in figure [ fenvar ] , as a function of coherent state number @xmath119 , for the same three points in parameter space for which we have investigated the energy convergence above . in all three cases the energy uncertainty decreases monotonically , and roughly as a power law @xmath271 .
the fact that the uncertainty clearly tends to zero as @xmath119 increases , shows that the trial wavefunction ( [ superpose ] ) converges to the true ground state as the number of coherent states is increased .
transverse component @xmath272 of the screening cloud , versus distance @xmath4 from the impurity , for @xmath273 , @xmath274 .
the black dashed curve is the ( fully converged ) result of a @xmath262 coherent state calculation .
the solid blue curve corresponds to the single coherent state result ( [ shcloud ] ) . in the spin - anisotropic limit @xmath208
, the single coherent state approximation is thus very accurate .
[ f3 ] ] transverse component @xmath272 versus distance @xmath4 , for @xmath275 , @xmath276 .
this parameter choice corresponds to an isotropic kondo coupling @xmath255 .
the black dashed line represents the result of a @xmath249 coherent state calculation , while solid curves represent results for @xmath277 and @xmath278 coherent states .
[ f4 ] ] next , we investigate the convergence of the screening cloud itself . in figure [ f3 ]
we show an instance of very rapid convergence in the case of strong spin - anisotropy , and in figure [ f4 ] an example where there is a noticeable change between @xmath207 and @xmath249 coherent states ( note also that this data set is one of the least converged ones included in this work ) . in both figures
we plot the @xmath245 transverse correlator @xmath279 , as we found this component to show the most dramatic change as @xmath119 is increased .
figure [ f3 ] corresponds to @xmath273 and @xmath280 ( strong spin - anisotropy ) , and the comparison to a @xmath281 coherent state calculation shows that the single coherent state ( silbey - harris ansatz ) is nearly exact in this limit .
the results in figure [ f4 ] were obtained at @xmath275 and @xmath276 , which corresponds to an isotropic kondo coupling @xmath255 , for @xmath282 and @xmath283 coherent states .
good convergence is clearly ensured by the computation with @xmath249 coherent states . in this subsection
, we give further evidence that ground state properties of the kondo model can be calculated accurately using the coherent state expansion . in particular , we focus here on several physical quantities that can be computed exactly via the bethe ansatz or related integrability techniques . as obtained via the multiple coherent state approximation ( dots ) and the exact bethe ansatz solution ( solid line ) , for @xmath274 .
[ f6 ] ] in figure [ f6 ] , we compare the multiple coherent state estimate for the ground state energy to results obtained via the exact bethe ansatz @xcite , at @xmath274 and various @xmath113 values .
the high energy cutoff @xmath59 corresponds to the parameter @xmath284 in ref . , and we used from this reference expression ( c.9 ) for the ground state energy and equation ( 8) for the kondo temperature @xmath285 , with the relationship between @xmath225 and @xmath284 given in ( c.8 ) .
although our calculation is variational , the energy is typically converged to about 0.1% ( see figure [ f2 ] ) .
one has to bear in mind however that , in the ultraviolet , the model for which the bethe ansatz yields the exact solution differs from the model we consider here .
the bethe ansatz result is only valid if all relevant energy scales in the problem are much smaller than the ultraviolet cut - off scale . for finite @xmath286 ,
the bethe ansatz expressions even present spurious divergences around @xmath287 , for @xmath288 as a result , exact agreement can only be expected at @xmath289 , and this explains why our variational result is slightly off the bethe ansatz result for small @xmath113 , although the variational approach presents better convergence in this regime .
in general , the numerical data agrees so closely with the bethe ansatz result that it is impossible to distinguish between errors resulting from truncation of the coherent state expansion at @xmath249 and errors due to ultraviolet differences between the models . between ground states at different and varying values of @xmath21 , but with the same fixed value of @xmath20 , versus the ratio @xmath290 . here
, the strongly spin - anisotropic regime is considered with @xmath273 , i.e. @xmath291 . the solid line shows the single coherent state formula ( [ rpa ] ) , whereas the dashed line is the exact analytical result ( [ luk1 ] ) .
the dotted line is the small @xmath113 , small @xmath193 approximation ( [ luk2]).[f7 ] ] at intermediate @xmath113 values .
open symbols in various colors show the @xmath281 coherent state results for @xmath292 ( i.e. @xmath293 ) for several values of @xmath21 and @xmath294 $ ] ( i.e. @xmath295 $ ] ) .
closed symbols show the @xmath249 coherent state results for @xmath296 ( i.e. @xmath297 ) for several values of @xmath21 and @xmath294 $ ] ( i.e. @xmath298 $ ] ) .
solid curves represent the exact analytical result ( [ luk1 ] ) .
the kondo temperature was taken from the approximate identification @xmath299 .
[ f7b ] ] at @xmath300 ( @xmath301 ) .
open symbols in various colors show the @xmath249 coherent state results for several values of @xmath21 and @xmath302 $ ] ( i.e. @xmath303 $ ] ) .
solid curves represent the exact analytical result ( [ luk1 ] ) . in the main panel ,
the improved renormalization group estimate ( [ tkren ] ) was used for @xmath285 . in the inset
, the less accurate value @xmath299 was used .
[ f7c ] ] another interesting quantity for which an exact analytical result has recently been obtained @xcite , and which gives some indirect information on the kondo cloud , is the overlap @xmath304 of two kondo ground states with different transverse exchange couplings , as a function of the kondo temperature ratio @xmath290 . here
@xmath305 and @xmath306 denote the full many - body ground states obtained at different @xmath21 but for the same @xmath20 , while @xmath307 and @xmath308 are the associated kondo temperatures . with the definition
@xmath309 $ ] , the exact result reads : @xmath310}{\sinh(z/2)}g_\alpha(z),\nonumber\\ & g_\alpha(z)=\exp\left[\int_0^\infty\frac{dt}{t}\frac{\sin^2(zt/\pi)}{\sinh(2t)\cosh(t ) } \frac{\sinh\left(\frac{\alpha t}{1-\alpha}\right)}{\sinh\left(\frac{t}{1-\alpha}\right)}\right].\label{luk1}\end{aligned}\ ] ] when comparing our numerical results to this analytical formula , a subtle issue arises .
the kondo temperature is certainly related to the inverse of the size of the screening cloud , but the exact relation may well involve an @xmath311 factor that is @xmath21 dependent .
we have therefore tried various definitions of the kondo temperature . for small to moderate @xmath113 , we find that @xmath312 , as defined in eq .
( [ asymptotics ] ) , with a @xmath21-independent proportionality constant , works well . indeed , in the silbey - harris regime , this correspondence is exact ( see below ) . for larger
@xmath113 however , this definition seems to incur a systematic error . for @xmath20 and @xmath21 relatively small ,
as is exemplified by the data we collected at @xmath300 , we have found it better to estimate the kondo scale from direct integration of the standard weak - coupling in @xmath21 and @xmath20 poor man s scaling equations : @xmath313 which lead to an expression of the kondo scale for the spin - anisotropic kondo model that is valid for @xmath314 and @xmath315 : @xmath316.\label{tkren}\ ] ] let us first consider the regime of small @xmath113 .
here we have seen that single coherent state results are already well - converged . in the single coherent state approximation ,
the overlap is given by : @xmath317,\label{rpa}\end{aligned}\ ] ] where @xmath318 $ ] .
for small @xmath113 , ref . quotes the `` semi - classical '' result , @xmath319,\label{luk2}\end{aligned}\ ] ] which can be obtained by expanding the exact result ( [ luk1 ] ) to first order in @xmath113 .
since @xmath320 grows linearly in @xmath19 for large @xmath19 , the semi - classical result can at best only be valid for @xmath19 sufficiently smaller than @xmath321 . referring back to the single coherent state result ( [ shxi ] ) for the correlation lengths , we see that if we make the identification @xmath312 , the single coherent state approximation ( [ rpa ] ) is nothing but a resummation of the small @xmath113 result in ref . , in which @xmath322 is calculated to second order in the impurity interaction , and then exponentiated . in figure [ f7 ]
we compare the single coherent state approximation to the exact result ( [ luk1 ] ) .
we see that , unlike the semiclassical formula ( [ luk2 ] ) , the single coherent state approximation remains valid up to large ratios of the kondo temperatures , because the ground state is well captured for any value of the kondo temperature ( provided @xmath113 is small enough ) .
we now turn to the regime of larger dissipation , and in figure [ f7b ] we show multiple coherent state results for the kondo overlap at @xmath293 and @xmath297 , again using the ( now approximate ) identification @xmath312 , together with the exact result ( [ luk1 ] ) .
data points of the same color were obtained by keeping @xmath21 fixed and varying @xmath323 . if two data points have different colors , they correspond to distinct @xmath21 _ and _ @xmath324 .
it is therefore already nontrivial that the differently colored points fall on the same curve , confirming the universality predicted by the exact analytical expression ( [ luk1 ] ) .
generally , good agreement with the analytical result ( [ luk1 ] ) is seen , with only a small systematic error pushing the numerical curves slightly below the analytical ones , which increases as @xmath113 increases .
a small part of the error is likely due to an error of a few percent in the numerical estimation of @xmath239 , and to @xmath285 being only an order of magnitude or so less than the ultraviolet scale @xmath59 at large @xmath21 .
these errors would be reduced if we could use more coherent states and smaller @xmath21 . however , as pointed out earlier , the main discrepancy comes from the chosen definition of the kondo temperature in ref .
, which is not exactly equivalent to @xmath325 at larger @xmath113
. indeed , figure [ f7c ] shows the multiple coherent state results for the overlap at @xmath300 , using both the naive identification @xmath312 ( inset ) and the renormalization improved estimate ( [ tkren ] ) .
the latter choice leads clearly to a substantial reduction of the error .
our conclusion is that the actual kondo overlap @xmath326 is calculated quite accurately in the coherent state expansion , and that the ( modest ) observed errors are associated with the extraction procedure of the kondo temperature from the size of the cloud .
component of the screening cloud at the toulouse point @xmath327 .
five different values of @xmath100 ranging from @xmath328 to @xmath329 were used .
the symbols correspond to results obtained with the coherent state expansion ( truncated at @xmath249 terms ) for a choice of discrete positions , while the solid lines represent the exact result ( [ xtoul ] ) . also indicated for each curve is the associated kondo length @xmath239 , ranging from @xmath330 at @xmath331 to @xmath332 at @xmath333.[ftoul ] ] in this subsection provide strong evidence that the coherent state expansion ( [ superpose ] ) captures the full spatial structure of the kondo cloud , using an exact analytical result at the so - called toulouse point .
indeed , at @xmath334 or equivalently , @xmath327 , the kondo model is equivalent to a fermionic non - interacting resonant level model [ ] .
an exact result for the longitudinal @xmath245 component of the screening cloud can be obtained following the route set out in refs . and .
we review this calculation in appendix [ appd ] .
the result is given by the simple formula @xmath335 with @xmath214 as defined in eq .
( [ efl ] ) . from the point of view of the coherent state expansion
, there is nothing special about the toulouse point . in the coherent state expansion , @xmath336
is expressed as a linear combination of @xmath214 functions with different position dependent arguments , cf .
( [ abcd ] ) , and an infinite number of terms are required to approximate @xmath337 exactly . in figure
[ ftoul ] , we compare the exact expression to results obtained with the coherent state expansion , truncated at @xmath249 terms . for clarity , we show here results of the coherent state expansion only for a discrete set of @xmath4 values , because the complete curves would have completely covered the exact result .
we clearly find near perfect agreement between the coherent state expansion and the exact result .
since the coherent state expansion does not exploit any special features of toulouse point , we expect similar accuracy at a similar cost ( @xmath249 ) to be achievable at other values of @xmath113 .
having thoroughly established the accuracy of the coherent state approximation for the kondo ground state , we now proceed to investigate the physical features of the kondo screening cloud , both for isotropic and anisotropic regimes .
we first briefly recall the available analytical results regarding the isotropic screening cloud . in the isotropic case ( @xmath71 ) ,
the transverse components of the screening cloud equal twice the longitudinal components , i.e. @xmath338 for @xmath92 , both @xmath5 and @xmath6 are expected to be universal scaling functions @xmath339 here @xmath340 is independent of @xmath9 , and all parameter dependence is contained in the kondo length @xmath10 .
the kondo length is expected to be inversely proportional to the kondo temperature , but the exact relation may contain a @xmath9-dependent proportionality factor of order unity , as we discussed previously .
the following asymptotic results for @xmath341 and @xmath342 have been derived analytically @xcite : @xmath343 the regime of small @xmath344 is perturbatively accessible with a calculation using renormalization group techniques , while the large @xmath344 regime is treated using fermi liquid theory .
note that at small @xmath344 , the @xmath87-oscillatory component of the cloud dominates the @xmath245 component slightly .
this implies that the total correlation function @xmath345 oscillates between positive and negative values , with wavelength @xmath346 at small @xmath4 . in other words , close to the impurity , the spin correlations between the electron gas and the impurity alternate between being ferromagnetic and being antiferromagnetic , on the scale @xmath346 .
the crossover from slower than @xmath347 decay inside the cloud to @xmath348 decay at large @xmath4 is expected to occur at @xmath349 , i.e. @xmath350 . our goal in the next paragraph is to confirm these results for the isotropic cloud , and to compute the full crossover curve from the coherent state expansion . in a second step , we will examine how the cloud correlations change when the kondo couplings are not isotropic ( @xmath351 ) .
components , while the bottom panel represents the longitudinal ( black ) and transverse ( gray ) @xmath87 components .
the calculation was performed for the choice of kondo couplings @xmath352 and @xmath353 , which correspond to @xmath354 and @xmath355 .
the various curves can be identified from the fact that the crossover to faster @xmath235 decay occurs on an increasing length scale @xmath10 as @xmath113 is increased ( or equivalently @xmath9 is decreased ) . for @xmath356 convergence
was achieved with @xmath357 coherent states , while for the other values of @xmath113 , @xmath249 was required.[fi1],title="fig : " ] + components , while the bottom panel represents the longitudinal ( black ) and transverse ( gray ) @xmath87 components .
the calculation was performed for the choice of kondo couplings @xmath352 and @xmath353 , which correspond to @xmath354 and @xmath355 .
the various curves can be identified from the fact that the crossover to faster @xmath235 decay occurs on an increasing length scale @xmath10 as @xmath113 is increased ( or equivalently @xmath9 is decreased ) . for @xmath356 convergence
was achieved with @xmath357 coherent states , while for the other values of @xmath113 , @xmath249 was required.[fi1],title="fig : " ] as a first step , we present results in figure [ fi1 ] for the screening cloud calculated for several parameters on the isotropic line @xmath71 , or equivalently @xmath358 .
the correlation functions are plotted in units of @xmath59 on the vertical axis and in units of @xmath269 on the horizontal axis .
when considering the small @xmath344 regime numerically , one must remember that @xmath4 has to remain sufficiently larger than the short distance ultraviolet cut - off @xmath269 , since the limits @xmath227 and @xmath228 do not commute .
this restriction is implied whenever we consider the asymptotic @xmath359 limit .
the transverse and longitudinal components are plotted in the same panel , and we expect @xmath360 for both the @xmath23 and @xmath24 components due to strict spin - isotropy .
the numerical results of figure [ fi1 ] reveal this isotropy to a high degree , and this is a non - trivial check of our method , since rotational symmetry in the bosonic model is only emergent .
it is violated by the ultraviolet regularization ( see appendix [ appmap ] ) .
this fact also explains the small differences that are still visible between the transverse and longitudinal components . , in rescaled units @xmath361 on the vertical axis and @xmath10 on the horizontal axis .
the kondo length @xmath362 was estimated from eq .
( [ ea1]).[fi2],title="fig : " ] + , in rescaled units @xmath361 on the vertical axis and @xmath10 on the horizontal axis .
the kondo length @xmath362 was estimated from eq .
( [ ea1]).[fi2],title="fig : " ] in figure [ fi2 ] we plot the same data as in figure [ fi1 ] , but now in rescaled units , according to ( [ eqisoscaling ] ) .
we ignored small differences between the large distance behavior of the transverse and longitudinal components , and scaled all components with @xmath362 , where the kondo length @xmath239 was calculated using eq .
( [ ea1 ] ) .
we clearly see that it is possible to scale correlation functions calculated at different @xmath9 onto universal curves . as expected , we observe that the cross - over from slower than @xmath363 decay inside the cloud to @xmath235 decay at large @xmath4 occurs around @xmath364
. the precise behavior of the kondo length @xmath10 as a function of @xmath9 will be analyzed further in the next subsection . at this point
we note that it varies from @xmath365 at @xmath366 to @xmath367 at @xmath368 .
such a large variation in the spatial scale implies that the observed scaling is non - trivial and reflects the universality of the kondo problem . ,
and the gray curve represents the transverse component @xmath369 .
these two single curves were obtained by fitting high order polynomials through the rescaled data set of figure [ fi2 ] .
thinner straight lines indicate the pure power laws @xmath370 and @xmath371 .
[ fi3 ] ] in figure [ fi3 ] we compare the universal scaling functions for the @xmath245 and the @xmath87 components of the cloud , by plotting them on top of each other .
we determined the single universal scaling functions by fitting high order polynomials through the scaled data set of figure [ fi2 ] .
we clearly see that the @xmath87 component dominates the @xmath245 component at small @xmath344 , consistent with the known small @xmath4 asymptotics of eq .
[ eqisoasym ] . having
confirmed that the cloud displays the expected universal scaling for isotropic couplings , we move on to the general anisotropic case .
the existence of two independent couplings @xmath20 and @xmath21 , or equivalently @xmath113 and @xmath100 , implies that the universal scaling picture is less straightforward . to guide our investigation ,
let us review known results obtained by an improved poor man s scaling argument , applicable with greater generality in the anisotropic case .
( standard poor man s equations ( [ poorfloweq ] ) can only be trusted when both kondo exchange couplings are small.)@xcite in the language of the spin - boson model , it is known that to leading order in @xmath100 , but arbitrary @xmath113 , increasing the short - distance scale @xmath269 by @xmath372 is approximately equivalent to changing @xmath113 to @xmath373 and @xmath100 to @xmath374 , where @xmath375 integration of these flow equations yields scaling trajectories @xmath376 or equivalently , in the language of the kondo model @xmath377\nonumber\\ & = \mbox{constant}.\end{aligned}\ ] ] a few of these trajectories in the @xmath20 - @xmath21 plane are plotted in figure [ figflow ] .
we stress that the standard weak coupling renormalization equations ( [ poorfloweq ] ) can not be trusted in the regime where @xmath378 , and that is why we have to work non - perturbatively in @xmath113 .
it is also important to remember that these trajectories are only meaningful at @xmath21 sufficiently smaller than @xmath99 . always bearing this proviso in mind ,
the following statement holds : screening clouds associated with two kondo hamiltonians , whose parameters lie on the same scaling trajectory , can be scaled onto the same universal line shape . ) , that are correct to leading order in @xmath21 . in order to emphasize that only the small @xmath21 part of trajectories are to be trusted ,
dashed lines are used for @xmath379 .
[ figflow ] ] and we generally find that more than @xmath249 coherent states are required for a converged result . above the upper solid curve , @xmath380 and the screening cloud is poorly resolved because of non - universal ultraviolet effects . at the points included in the shaded region , we find that the screening cloud is nearly isotropic down to distances deep inside the cloud.[fa1 ] ] of longitudinal and transverse cloud components plotted against the scaled distance @xmath381 .
perfect isotropy corresponds to @xmath382 .
solid curves correspond to @xmath275 and various values of @xmath100 ranging from @xmath383 to @xmath384 .
dashed curves correspond to @xmath356 and various values of @xmath100 ranging from @xmath385 to @xmath386 .
the kondo length @xmath239 was calculated for each curve from eq .
( [ ea1 ] ) . for the set of data at @xmath275
, @xmath239 varies from @xmath387 to @xmath388 , while for @xmath356 , @xmath239 varies from @xmath389 to @xmath390 .
[ figratio ] ] ( top panel ) and at @xmath297 ( bottom panel ) , to the universal isotropic line shapes ( full and dashed black lines ) .
the curves were obtained by fitting data sets such as those displayed in the main panels of fig .
[ f8 ] to polynomials of high degree . on the vertical axis ,
the longitudinal components are plotted in units of @xmath325 , while the transverse components are plotted in units of @xmath391 .
the isotropic curves are those of figure [ fi3].[f3comp],title="fig : " ] ( top panel ) and at @xmath297 ( bottom panel ) , to the universal isotropic line shapes ( full and dashed black lines ) .
the curves were obtained by fitting data sets such as those displayed in the main panels of fig .
[ f8 ] to polynomials of high degree . on the vertical axis ,
the longitudinal components are plotted in units of @xmath325 , while the transverse components are plotted in units of @xmath391 .
the isotropic curves are those of figure [ fi3].[f3comp],title="fig : " ] at @xmath293 ( black ) and on the isotropic line ( gray ) , as shown by full lines ( dashed lines are the respective envelope functions @xmath392 ) .
both curves were normalized using their magnitude at @xmath393 .
for the purpose of showing the oscillations , we took @xmath394.[figfriedel ] ] below , our work will be guided by two qualitative features of figure [ figflow ] .
the first is that the flow trajectories intersect the @xmath20-axis at @xmath395 . for large @xmath20 , the parts of the trajectories that may reasonably be trusted , run nearly parallel to the @xmath21 axis .
going beyond the small @xmath21 regime of the figure , we know that , at the toulouse point @xmath334 , the scaling trajectory runs exactly parallel to the @xmath21 axis .
( see appendix [ appd ] . )
universality at fixed @xmath20 is also an obvious feature of the exact analytical expression for the kondo overlap ( [ luk1 ] ) . for @xmath396 ( @xmath397 )
, we will therefore try to scale clouds at fixed @xmath20 , but different @xmath21 , onto each other .
the second pertinent feature of figure [ figflow ] is that for @xmath398 , trajectories are approximately hyperbolas @xmath399 . in other words ,
the isotropic line @xmath400 is an attractor for the renormalization flow .
this implies that at sufficiently large distances @xmath401 , the cloud must tend to the isotropic cloud .
however , for @xmath402 , the poor man s scaling picture of figure [ figflow ] indicates pronounced anisotropic behavior .
this short distance region corresponds to the scales that have to be integrated out for an anisotropic point @xmath403 to flow close to the isotropic line .
note , furthermore that the above discussion only applies to points that are in the general vicinity of the isotropic line , and are associated with a kondo temperature sufficiently lower than @xmath59
. if a point is too far from the isotropic line to start with , or if the initial kondo temperature is too large , the size of the cloud renormalizes down to the ultraviolet scale , before the renormalized couplings become isotropic . at this point
the notion of flow trajectories in a two - dimensional parameter space breaks down , and kondo physics mixes with ultraviolet physics . based on these observations
, we will compare the cloud calculated at @xmath404 ( i.e. reasonably close to the isotropic line and with a decent kondo length ) with the isotropic cloud .
we start the presentation of our results in the regime of small @xmath113 ( large @xmath20 ) .
this regime does not overlap with the isotropic kondo regime , because the size of the screening cloud flows to the short distance cut - off @xmath269 before the renormalized couplings come close to the isotropic line . for small @xmath113 , the single coherent state ( silbey - harris ) approximation is accurate ( see figure [ f3 ] ) , and
we have derived simple analytical expressions ( [ shcloud ] ) for the cloud correlation functions , that reveal the following universality . by appropriately rescaling the four cloud components and
the coordinate @xmath4 , clouds calculated at different @xmath405 but fixed @xmath20 ( i.e. fixed @xmath113 ) have the same line shape : for the longitudinal components @xmath406 , @xmath407 one defines @xmath408 and finds from the silbey - harris result the simple analytical formulas : @xmath409.\label{uncomps1}\end{aligned}\ ] ] in the small @xmath113 regime , the two transverse components @xmath410 , @xmath407 , are equal .
one defines @xmath411 and finds the analytical expression : @xmath412.\label{uncomps2}\ ] ] the expression for @xmath413 could suggest a further rescaling @xmath414 that would lead to an @xmath113 independent line shape for @xmath336 .
however , one can not simultaneously get rid of the @xmath113 dependence in the other components of the cloud .
furthermore , it turns out that @xmath415 is no longer @xmath113-independent beyond @xmath416 ( as shown by our numerical simulations below ) . the two pertinent facts to emerge from this discussion are the following .
firstly , whereas the magnitudes of the longitudinal and transverse components of the cloud have to be rescaled by different inverse lengths @xmath325 and @xmath417 , the coordinate @xmath4 is always rescaled by @xmath239 . in other words ,
the longitudinal and transverse clouds have different characteristic magnitudes @xmath325 and @xmath417 , but the same characteristic size @xmath239 .
secondly , the universal scaling curves depend in general on @xmath113 .
thus , as expected from poor man s scaling , clouds at the same @xmath20 but different @xmath21 have the same shape , whereas clouds at different @xmath20 in general have different shapes . the differences in scaling behavior between the small @xmath113 and isotropic clouds are thrown into sharp relief when one considers the small @xmath4 asymptotic form of the cloud correlators . at @xmath418 , the small @xmath113 results ( [ uncomps1 ] ) and ( [ uncomps2 ] ) lead to asymptotic behavior : @xmath419 the @xmath420 component diverges much more slowly in the @xmath421 limit than is the case at isotropic couplings ( @xmath422 vs. @xmath423 ) , whereas the @xmath424 component diverges more rapidly than in the isotropic case ( @xmath425 vs. @xmath426 ) .
thus , the alternation between ferro- and antiferromagnetic correlations close to the impurity is enhanced in the longitudinal direction .
this is not surprising , as small @xmath113 implies @xmath41 .
furthermore we see that the small @xmath4 asymptotic behavior of the transverse cloud is explicitly @xmath113-dependent , displaying power - law behavior with an exponent @xmath427 , implying a divergence as @xmath428 that is closer to @xmath235 , than to @xmath363 .
the next obvious question is how the small @xmath113 scaling picture evolves as @xmath113 increases . in order to investigate this , we calculated the cloud correlators at the points in parameter space shown in figure [ fa1 ] .
the most obvious feature of the data , is that the cloud becomes more and more isotropic as @xmath113 increases .
this is revealed in figure [ roadtoiso ] , where we plot @xmath429 and @xmath430 on top of each other for respectively @xmath293 , @xmath431 , @xmath297 , and @xmath300 , in separate panels .
each panel contains curves for several values of @xmath21 . at @xmath293 ,
the transverse and longitudinal clouds are very different . as @xmath113 increases ,
the differences become smaller . by the time we reach @xmath300 , the cloud is isotropic to a high degree of accuracy for all values of @xmath21 considered .
the behavior of the @xmath432 and @xmath433 components of the cloud ( not shown ) present very similar behavior .
we find similarly isotropic clouds at all the points in figure [ fa1 ] inside the shaded region .
poor man s scaling suggests that for anisotropic exchange couplings , there is a scale @xmath434 below which isotropy breaks down . to investigate this
, we plot the ratio @xmath435 as a function of @xmath381 ( figure [ figratio ] ) .
we do so for the data collected at @xmath356 and for @xmath275 . in both cases we see that @xmath435 reaches a plateau that is within 10% of unity .
this is the same degree of isotropy as we obtain on the isotropic line itself . in both cases ,
the plateau is reached at a distance that is a small fraction of the kondo length @xmath238 . in other words
, isotropy extends deep inside the cloud .
the ratio @xmath436 between the scale where isotropy sets in , and the kondo length , is more or less constant at fixed @xmath113 , while the kondo length itself varies by an order of magnitude . from the figure one can read off that @xmath437 for @xmath356 and @xmath438 for @xmath275 .
next , we investigate the precise scaling behavior of the data . for @xmath439 ,
we take our cue from the small @xmath113 results we obtained above , together with the understanding from poor man s scaling , and we thus scale correlation functions according to the ansatz ( [ ansatz1 ] ) and ( [ ansatz2 ] ) . in fig .
[ f8 ] , we show raw and scaled results for @xmath297 . similar figures for @xmath293 and @xmath440 can be found in the supplementary material.@xcite for all values of @xmath113 that we considered , the data nicely scale onto single universal curves .
we conclude that for practical purposes , scaling ansatz of the form ( [ ansatz1 ] ) and ( [ ansatz2 ] ) hold for all values of @xmath441 .
these universal scaling curves are @xmath113-dependent and differ from the isotropic scaling curves .
this can clearly be seen in figure [ f3comp ] , where we compare the universal line shapes of the scaled cloud at @xmath293 ( top panel ) and at @xmath297 ( bottom panel ) to the universal isotropic line shapes obtained in sec .
[ secisocase ] .
although the single coherent state ( silbey - harris ) approximation underestimates the kondo length @xmath239 by @xmath442 at @xmath293 , the simple equations ( [ uncomps1]-[uncomps2 ] ) still predict the universal line shape of the screening cloud very well . inside the cloud ( @xmath443 ) , these universal curves are very different from the universal isotropic curves . as @xmath113 increases , deviations from the silbey - harris line - shape set in , and scaling curves start to resemble the isotropic curves more closely , as can be seen for @xmath297 in the bottom panel of figure [ f3comp ] .
a prominent feature of the anisotropic scaling curves is that , at small distances , there is a stronger suppression of the @xmath444 component over the @xmath445 component , than in the isotropic case . as a result ,
the alternation from ferro- to antiferromagnetic correlations between the impurity and the electron gas inside the cloud is enhanced in the longitudinal direction .
this is illustrated in figure [ figfriedel ] , where we compare the full longitudinal cloud @xmath446 calculated at @xmath293 with the result on the isotropic line . ) in figure [ fa1 ] .
insets show the raw data , while the main panels show data in units of @xmath239 on the horizontal axis and units of @xmath325 on the vertical axis .
the top panel contains the transverse and longitudinal @xmath447 components plotted together .
the bottom panel contains the transverse and longitudinal @xmath87 components.[fall],title="fig : " ] ) in figure [ fa1 ] .
insets show the raw data , while the main panels show data in units of @xmath239 on the horizontal axis and units of @xmath325 on the vertical axis .
the top panel contains the transverse and longitudinal @xmath447 components plotted together .
the bottom panel contains the transverse and longitudinal @xmath87 components.[fall],title="fig : " ] , i.e. clouds calculated at all the points inside the shaded region in figure [ fa1 ] .
these curves were obtained as the best fit of a high order polinomial through the scaled data of figure [ fall ] .
the dashed black curves are the universal scaling curves inferred using only clouds calculated at perfectly isotropic couplings ( see figure [ fi3 ] ) .
[ fallviso],title="fig : " ] + to complete the scaling analysis , we consider the screening cloud at @xmath448 ( the shaded region in figure [ fa1 ] ) . in this region , we have seen that the cloud is approximately isotropic . in figure
[ fall ] , we therefore plot the transverse and longitudinal components of the cloud on the same graph .
insets show raw results in units of @xmath269 on the horizontal and @xmath59 on the vertical axes .
the main panels show data in rescaled units of @xmath239 on the horizontal and @xmath325 on the vertical axes .
we see that the scaled data collapse very well onto single scaling curves . as before , the kondo length @xmath239 was calculated using eq .
( [ ea1 ] ) , i.e. no fitting parameters were used to obtain the high degree of collapse . it is instructive to compare the main panels of figure [ fall ] for all the data collected , to figure [ fi2 ] , that shows scaled data only for the subset of points that lie on the isotropic line @xmath71 .
we see the same degree of collapse onto single curves in both figures . in other words , within the numerical accuracy of our calculation , the universal line shape of the screening cloud is the same for all points in the shaded region of figure [ fa1 ] . to further confirm this conclusion ,
we fitted a high order polynomial through the complete scaled data sets of figure [ fall ] . in figure [ fallviso ]
we compare this fit to the universal scaling curves we obtained previously by only considering the cloud on the isotropic line ( figure [ fi3 ] ) .
we see nearly perfect agreement .
of course , the screening cloud at an anisotropic value of the exchange couplings , @xmath449 , only follows this universal line shape for distances larger than @xmath434 , the scale at which isotropy sets in .
however , as we have seen in figure [ figratio ] , there is a sizable region around the isotropic line where isotropy already sets in deep inside the cloud . as a function of @xmath21 for several @xmath113 values .
the solid lines shows @xmath450 , with @xmath285 the standard poor man s scaling estimate ( [ tkren ] ) and @xmath451 a fitting parameter . the same value of @xmath452 we used for all five curves.[fklen ] ] finally , we plot in figure [ fklen ] the kondo length @xmath239 , which corresponds to the characteristic size of the cloud , for some of the data we have collected .
the large range over which @xmath239 varies for each value of @xmath113 , confirms that the scaling behavior we see is non - trivial .
we compare the calculated value of @xmath239 to @xmath453 , with @xmath285 the standard poor man s scaling estimate ( [ tkren ] ) of the kondo temperature . with a logarithmic scale on the vertical axis , which hides mismatches by factors of order unity
, we find good agreement for the isotropic and the @xmath300 data sets . that these sets yield the best agreement is expected , as they are closer to the point @xmath454 in parameter space than the other sets , and the version of poor man s scaling that leads to the estimate ( [ tkren ] ) for @xmath285 assumes small exchange couplings .
this paper has provided a very extensive study of the kondo screening cloud for a wide range of parameter values , including the spin - isotropic and strongly spin - anisotropic regimes .
methodologically , we derived simple but controlled analytical expressions in the case of strong anisotropy , and developed an original and very powerful numerical technique to tackle the problem in its complete generality .
this allowed us to investigate the @xmath245 and @xmath87 component of the kondo cloud correlator separately , both for the longitudinal and transverse response .
in addition , we have examined , again both analytically and numerically , the structure of kondo overlaps introduced in recent works.@xcite our main results concern the universal scaling of correlations between the impurity spin and the electron spin density in the anisotropic kondo model .
they can be summarized as follows . at large @xmath4 , transverse correlators equal @xmath455 while longitudinal correlators equal @xmath456 ( here @xmath457 refer to respectively the forward or backscattering component of the correlator ) . in general , the two emergent length scales @xmath238 and @xmath239 are not equal , and we find the following universal scaling behavior : @xmath458 the scaling curves are @xmath21-independent , but remain @xmath113-dependent ( and therefore @xmath20-dependent ) . as can be seen from the scaling equations , the two lengths @xmath238 and @xmath239 have fundamentally different roles .
the length @xmath239 sets the size of the screening cloud , whereas the ratio @xmath459 sets the relative magnitude of transverse correlations , compared to longitudinal correlations .
as @xmath20 decreases , the explicit @xmath20 dependence of the scaling curves becomes weaker and weaker . for @xmath460
, we find no discernible @xmath20-dependence anymore , and the universal scaling curves become those of the isotropic model . in this region of parameter space , we find that isotropy already sets in deep inside the cloud .
we managed to calculate the universal scaling curves down to distances @xmath461 or less .
this far inside the cloud , the backscattering ( @xmath87 ) component of the longitudinal correlator dominates the forward scattering ( @xmath245 ) component . inside the cloud , correlations between the impurity spin and the electron spin density therefore alternate with wave vector @xmath87 between being ferro- and antiferromagnetic .
the effect is more pronounced in the regime @xmath41 , but we can still clearly resolve it when the cloud becomes isotropic . in previous numerical studies , results were obtained using the numerical renormalization group ( nrg ) .
this required solving a two - impurity problem , where the position of the second ( fictitious ) impurity represents the point @xmath4 where the correlator is evaluated .
an independent two channel kondo problem must be solved by nrg at every position where the correlator is evaluated , making the calculation computationally expensive .
in addition , it proved difficult to resolve in nrg the dominance of the backscattering components of the cloud over the forward scattering components inside the cloud .
we have shown that the coherent state expansion , the method we employed in this article , complements previous numerical renormalization group studies , and even offers simple analytical insights in the regime of strong spin - anisotropy .
thanks to a technical improvement that dramatically reduces the number of variational parameters , it offers excellent accuracy at a reduced computational cost , and does not require one to discretize the bath degrees of freedom . yet , we note that the numerical renormalization group was successfully used to study finite temperature @xcite and time - dependent@xcite features of the kondo cloud .
we envisage future work to extend the coherent state expansion into these domains as well .
in this appendix , we review the mapping from the kondo hamiltonian ( [ h2 ] ) to the spin - boson hamiltonian ( [ sbham ] ) .
the first step is to express the ( even ) fermionic degrees of freedom in terms of bosonic ones @xmath462 , @xmath463 , @xmath464 with @xmath465=\delta_{qq'}\delta_{\sigma\sigma'}$ ] . for this purpose
, we invoke the bosonization identities @xmath466 and @xmath467 here @xmath468 denotes the unitary klein factor operator associated with increasing the number of @xmath47 electrons by one .
we separate spin and charge degrees of freedom by defining : @xmath469 we also define new su(2 ) operators , in which fermion number and impurity degrees of freedom are mixed .
@xmath470 in the bosonic representation , the hamiltonian reads @xmath471,\label{bosrep}\end{aligned}\ ] ] where @xmath472 we note that spin ( @xmath73 ) and charge ( @xmath473 ) degrees of freedom are decoupled , and that only spin degrees of freedom couple to the impurity .
we restrict the hamiltonian to the vacuum of the charge sector and drop the @xmath474 terms in future expressions .
finally , we bring the hamiltonian into the standard form of the ( unbiased ) spin - boson model , via a unitary transform @xmath475 under the action of @xmath141 , @xmath476 so that @xmath477 , where @xmath478 is given by ( [ sbham ] ) .
in this appendix , we transform the fermionic cloud correlators ( [ clouddef ] ) onto their bosonic counterparts ( [ bosoncloud ] ) . to do so , we represent the electron spin density in terms of bosonic degrees of freedom , and subsequently perform the unitary transformation of ( [ uni ] ) on these operators . to this end , it is useful to note from ( [ psi ] ) the following identities : @xmath479 \simeq e^{ik_f x}\psi_\sigma(x)+ e^{-ik_f x}\psi_\sigma(-x),\nonumber\\ & \frac{1}{\sqrt{2}}\left[\tilde \psi_\sigma(x)-\tilde \psi_\sigma(-x)\right ] \simeq
e^{ik_f x}\bar \psi_\sigma(x)- e^{-ik_f x}\bar \psi_\sigma(-x).\label{slow}\end{aligned}\ ] ] the ` approximately equal ' signs in these equations indicate that the operators on the right hand side represent versions of the point operators on the left , that have been broadened by an amount @xmath480 , inversely proportional to the ultraviolet cut - off . now consider the expectation value @xmath481 , which , from ( [ clouddef ] ) , can also be written as @xmath482 because the ground state has good spatial parity , @xmath481 is an even function , and we may write @xmath483\right>_{\rm k}.\ ] ] the last expression is manipulated to become @xmath484\left[\tilde\psi_\downarrow(x)+\tilde\psi_\downarrow(-x)\right]\right>_{\rm k } + \underbrace{\left<\sigma^-\left[\tilde\psi^\dagger_\uparrow(x)-\tilde\psi^\dagger_\uparrow(-x ) \right]\left[\tilde\psi_\downarrow(x)-\tilde\psi_\downarrow(-x)\right]\right>_{\rm k}}_{=0}\label{xp}\right\}.\end{aligned}\ ] ] the term marked with the underbrace is zero because of the following reason : @xmath485 creates an electron in an odd parity single particle orbital , and these orbitals do not couple to the impurity , so that : @xmath486\left[\tilde\psi_\downarrow(x)-\tilde\psi_\downarrow(-x)\right]\right>_{\rm k } = \left<\sigma^-\right>_{\rm k}\left<\left[\tilde\psi^\dagger_\uparrow(x)-\tilde\psi^\dagger_\uparrow(-x ) \right]\left[\tilde\psi_\downarrow(x)-\tilde\psi_\downarrow(-x)\right]\right>_{\rm k}.\ ] ] since the antiferromagnetic kondo ground state is a singlet , @xmath487 . substituting from ( [ slow ] ) into the first line of ( [ xp ] ) , we obtain @xmath488\right>_{\rm k}.\end{aligned}\ ] ] performing a similar calculation for @xmath489 , we find @xmath490\right>_{\rm k},\end{aligned}\ ] ] where @xmath491 and @xmath492 .
we finally manipulate the above expressions as follows : 1 .
the fermion operators are bosonized .
we apply the unitary transformation that maps the kondo hamiltonian onto the spin - boson model . in the resulting expressions , expectation values
are with respect to the ground state of the spin - boson model , indicated without a subscript .
3 . we normal order the bosonic operators . in the resulting expressions , the operators @xmath493 and @xmath473 , are replaced by zeros , since @xmath473 annihilates the ground state . finally , we take the large system limit , which allows us to make replacements such as @xmath494 .
the resulting expressions are those presented in the main text ( [ bosoncloud ] ) .
the integrals involved in calculating the cloud are of the form : @xmath495 where the function @xmath496 goes to zero as least as fast as @xmath497 for large @xmath128
. furthermore , the only non - analyticities that @xmath496 possesses in the complex @xmath128 plane are ( first order ) poles at @xmath160 , @xmath498 .
these integrals can be performed by extending the method employed in sec .
[ secen ] , where the energy was calculated .
define @xmath499 and @xmath500 . integrating around the shaded area @xmath224
, whose azimuthal boundary is understood to lie at @xmath501 in figure [ con2 ] , we obtain @xmath502.\end{aligned}\ ] ] .[con2 ] ] the integral on the right hand side on the first line is of precisely the same form as those considered in sec .
[ secen ] .
we also note that the summation on the second line vanishes if there are no poles @xmath160 with positive real parts . using the results of sec .
[ secen ] , we obtain : @xmath503\nonumber\\ & + 2\pi i\sum_{\omega_n\in a}{\rm res}\left[e^{-zq}r(q),\omega_n\right].\end{aligned}\ ] ]
in this appendix , we review the mapping between the kondo model at the toulouse point @xmath327 ( i.e. @xmath334 ) , and the non - interacting fermionic resonant level model , following ref . .
we also derive an exact expression for the longitudinal @xmath245 component of the screening cloud .
the derivation takes the same route as that set out in refs . and .
the starting point is the bosonic representation ( [ bosrep ] ) of the kondo hamiltonian .
the fermionic non - interacting resonant level @xmath504 is obtained , by applying the unitary transform @xmath505 , where @xmath506\right\}.\ ] ] the bosonic field @xmath507 is defined in eq .
( [ appphidef ] ) , and the operator @xmath508 is the @xmath19-component of the total spin of the electron gas .
new degrees of freedom @xmath509\frac{f_\uparrow^\dagger
f_\downarrow}{\sqrt{2\pi a}}\nonumber\\ & = & \frac{1}{\sqrt{l}}\sum_k e^{-ikx}e^{-a|k|/2}\bar c_k^\dagger\nonumber\\ d^\dagger&=&\exp\left[i\pi\left(n_-\frac{1}{2}\right)\right]\sigma^+,\end{aligned}\ ] ] emerge , that obey the usual fermionic commutation relations . here
@xmath510 with @xmath511 defined in eq . ( [ phixdef ] ) , and @xmath512 denote klein factors . in terms of these new fermions
, @xmath504 reads @xmath513.\ ] ] at the toulouse point , the unitary transformation @xmath514 maps the ground state of the kondo hamiltonian onto the ground state of @xmath504 . since @xmath504 is a quadratic hamiltonian ,
ground state correlation functions can be calculated straight - forwardly for transformed observables , provided they are expressed simply in terms of the new fermionic degrees of freedom . here
we focus on the longitudinal @xmath245 component @xmath336 for which this is easily accomplished . applying the unitary mapping to the @xmath19-components of the impurity spin and the electron spin density operators
, one finds @xmath515\right>\nonumber\\ & -(2-\sqrt{2})\frac{a}{x^2+a^2},\end{aligned}\ ] ] where the expectation value on the right hand side of the first line is with respect to the ground state of @xmath504 .
this expression is further simplified using wick s theorem and noting that the occupation probability for the resonant level is one half , i.e. @xmath516 . in terms of the original degrees of freedom
, this corresponds to @xmath517 which is guaranteed by the singlet nature of the kondo ground state .
working directly with @xmath504 , this follows from particle - hole symmetry .
( note that the energy of the resonant level co - incides with the fermi energy . )
to be more precise , @xmath504 is invariant under the combined action of particle - hole and spatial inversion . owing to the same symmetry , @xmath518 . putting everything together ,
one obtains @xmath519 the expectation value @xmath520 can be calculated from the known single particle green s functions of the resonant level model . in the thermodynamic limit ( @xmath167 )
the exact result is @xmath521,\label{cor1}\end{aligned}\ ] ] where @xmath522 is a regularized step function @xmath523-e^{-\omega a}\frac{i\gamma(0,\omega(ix - a))}{2 \pi},\end{aligned}\ ] ] with @xmath524 , and @xmath525 is the retarded resonant level self - energy @xmath526.\end{aligned}\ ] ] in general , the integral ( [ cor1 ] ) has to be evaluated numerically .
however , for @xmath527 , the following approximations are valid , which yields an analytical result . as a function of ( negative ) @xmath528 , the regularized step function @xmath522 is suppressed as @xmath529 for @xmath191 . as a result
, the integrand of ( [ cor1 ] ) is strongly suppressed for @xmath530 . as a function of @xmath4 , the regularized step function
is smoothed at a scale @xmath531 . for @xmath532 and @xmath527 ,
the regularized step function can be replaced with the sharp step function @xmath533 . for @xmath92 , the rapidly oscillating factor
@xmath534 then cuts off the integral at @xmath535 , so that the self - energy @xmath525 , that varies on the scale of @xmath59 , can be evaluated at @xmath536 @xmath537 for @xmath527 , one then finds @xmath538 where @xmath231 is defined in eq .
( [ efl ] ) .
we substitute this into ( [ toulcloud1 ] ) and also drop the second term in that equation , which is valid for @xmath527 , to obtain @xmath539 the kondo length @xmath239 is defined as in ( [ asymptotics ] ) in the main text , i.e. @xmath540 .
this yields @xmath541 and a universal scaling function ( cf .
( [ ansatz1 ] ) ) @xmath542 this work is based on research supported in part by the national research foundation of south africa ( grant number 90657 ) . 99 a. c. hewson , _ the kondo problem to heavy fermions _ ( cambridge university press , cambridge , uk , 1993 ) .
i. affleck , in _ perspectives of mesoscopic physics : dedicated to yoseph imry s 70th birthday _ , a. aharony and o. entin - wohlman ( eds . ) ( world scientific , singapore , 2010 ) .
v. barzykin and i. affleck , phys .
76 * , 4959 ( 1996 ) .
j. e. gubernatis , j. e. hirsch , and d. j. scalapino , phys .
b * 35 * , 8478 ( 1987 ) .
l. borda , phys .
b * 75 * , 041307(r ) ( 2007 ) . b. lechtenberg and f. anders , phys . rev .
b * 90 * , 045117 , ( 2014 ) .
i. affleck and p. simon , phys .
lett . * 86 * , 2854 ( 2001 ) .
p. s. cornaglia and c. a. balseiro , phys .
lett . * 90 * , 216801 ( 2003 ) . t. hand , j. kroha , and h. monien , phys
lett . * 97 * , 136604 ( 2006 ) .
r. g. pereira , n. laflorencie , i. affleck , and b. i. halperin , phys .
b * 77 * , 125327 ( 2008 ) . j. park , s .- s .
b. lee , y. oreg , and h .- s .
sim , phys .
* 110 * , 246603 ( 2013 ) .
y. nishida , phys .
* 111 * , 135301 ( 2013 ) .
j. bauer , c. salomon , and e. demler , phys .
lett . * 111 * , 215304 ( 2013 ) .
i. affleck , l. borda , and h. saleur , phys .
b * 77 * , 180404 ( 2008 ) .
a. k. mitchell , m. becker , and r. bulla , phys .
b 84 , 115120 ( 2011 ) .
a. holzner , i. p. mcculloch , u. schollwck , j. von delft , and f. heidrich - meisner , phys .
b * 80 * , 205114 ( 2009 ) . c. a. bsser , g. b. martins , l. costa ribeiro , e. vernek , e. v. anda , and e. dagotto , phys . rev .
b * 81 * , 045111 ( 2010 ) . i. snyman and s. florens , phys . rev .
b * 92 * , 085131 ( 2015 ) .
m. goldstein , m. h. devoret , m. houzet , and l. i. glazman , phys .
lett . * 110 * , 017002 ( 2013 ) .
s. chakravarty , phys .
lett . * 49 * , 681 ( 1982 ) .
s. l. lukyanov , h. saleur , j. l. jacobsen , and r. vasseur , phys .
lett . * 114 * , 080601 ( 2015 ) .
s. bera , s. florens , h. u. baranger , n. roch , a. nazir , and a. w. chin , phys .
b * 89 * , 121108(r ) ( 2014 ) .
s. bera , a. nazir , a. w. chin , h. u. baranger , and s. florens , phys .
b * 90 * , 075110 ( 2014 ) .
a. j. leggett , s. chakravarty , a. t. dorsey , m. p. a. fisher , a. garg , and w. zwerger , rev .
phys . * 59 * , 1 ( 1987 ) .
u. weiss , _ quantum dissipative systems _ ( world scientific , 1993 ) .
f. guinea , v. hakim , and a. muramatsu , phys .
b * 32 * , 4410 ( 1985 ) .
g. kotliar and q. si , phys .
b * 53 * , 12373 , ( 1996 ) . t. a. costi and g. zarnd , phys . rev .
b * 59 * , 12398 , ( 1999 ) . v. j. emery and a. luther , phys . rev .
* 26 * , 1547 ( 1971 ) .
r. silbey and r. a. harris , j. chem . phys . *
80 * , 2615 ( 1984 ) ; r. a. harris and r. silbey , j. chem . phys . * 83 * , 1069 ( 1985 ) .
p. w. anderson , phys .
164 * , 352 ( 1967 ) . p. w. anderson , in _ many - body problem _ , c. de witt and r. balian ( eds . ) ( gordon and breach , new york , 1968 ) .
g. bergmann and l. zhang , phys .
b * 76 * , 064401 ( 2007 ) .
k. yosida , phys . rev .
* 147 * , 223 ( 1966 ) .
a. okiji , prog .
theor . phys . *
36 * , 712 ( 1966 ) .
k. yosida and a. yoshimori , prog .
. phys . * 42 * , 753 , ( 1969 ) .
k. e. cahill , phys . rev . * 138 * , b1566 , ( 1965 ) .
f. w. j. olver , d. w. lozier , r. f. boisvert , and c. w. clark ( eds . ) , _ nist handbook of mathematical functions _
( cambridge university press , 2010 ) .
a. corana , m. marchesi , and s. ridella , acm trans .
* 13 * , 262 ( 1987 ) .
w. l. goffe , g. d. ferrier , and j. rogers , j. econ .
* 60 * , 65 ( 1994 ) .
k. le hur , ann .
physics ( ny ) , * 323 * , 2208 , ( 2008 ) g. zarnd and j. von delft , phys .
b * 61 * , 6918 , ( 2000 ) .
t. posske , c - x liu , j. c. budich , and b. trauzettel , phys .
lett . * 110 * , 016602 , ( 2013 ) .
t. posske and b. trauzettel , phys .
b * 89 * , 075108 , ( 2014 ) .
* supplementary information for `` universal spatial correlations in the anisotropic kondo screening cloud : analytical insights and numerically exact results from a coherent state expansion '' *
in the main text , we showed extensive data at @xmath297 ( i.e. @xmath543 ) demonstrating that cloud correlation functions calculated at different values of @xmath544 scaled onto the same universal functions ( figure [ f8 ] ) . here , we show for completeness data that demonstrate scaling at @xmath293 ( figure [ fapp1 ] ) , @xmath431 ( figure [ fapp2 ] ) , and @xmath300 ( figure [ fapp3 ] ) .
( @xmath545 ) .
raw data is shown in the insets , with different curves correspond to five different values of @xmath100 collected in the interval @xmath546 $ ] ( i.e. @xmath547 $ ] ) .
[ fapp1],title="fig : " ] ( @xmath545 ) .
raw data is shown in the insets , with different curves correspond to five different values of @xmath100 collected in the interval @xmath546 $ ] ( i.e. @xmath547 $ ] ) .
[ fapp1],title="fig : " ] ( @xmath545 ) .
raw data is shown in the insets , with different curves correspond to five different values of @xmath100 collected in the interval @xmath546 $ ] ( i.e. @xmath547 $ ] ) .
[ fapp1],title="fig : " ] ( @xmath545 ) .
raw data is shown in the insets , with different curves correspond to five different values of @xmath100 collected in the interval @xmath546 $ ] ( i.e. @xmath547 $ ] ) .
[ fapp1],title="fig : " ] ( @xmath548 ) .
raw data is shown in the insets , with different curves correspond to six different values of @xmath100 collected in the interval @xmath549 $ ] ( i.e. @xmath550$]).[fapp2],title="fig : " ] ( @xmath548 ) .
raw data is shown in the insets , with different curves correspond to six different values of @xmath100 collected in the interval @xmath549 $ ] ( i.e. @xmath550$]).[fapp2],title="fig : " ] ( @xmath548 ) . raw data is shown in the insets , with different curves correspond to six different values of @xmath100 collected in the interval @xmath549 $ ] ( i.e. @xmath550$]).[fapp2],title="fig : " ] ( @xmath548 ) .
raw data is shown in the insets , with different curves correspond to six different values of @xmath100 collected in the interval @xmath549 $ ] ( i.e. @xmath550$]).[fapp2],title="fig : " ] ( @xmath301 ) . raw data is shown in the insets , with different curves correspond to eight different values of @xmath100 collected in the interval @xmath303 $ ] ( i.e. @xmath551$]).[fapp3],title="fig : " ] ( @xmath301 ) .
raw data is shown in the insets , with different curves correspond to eight different values of @xmath100 collected in the interval @xmath303 $ ] ( i.e. @xmath551$]).[fapp3],title="fig : " ] ( @xmath301 ) . raw data is shown in the insets , with different curves correspond to eight different values of @xmath100 collected in the interval @xmath303 $ ] ( i.e. @xmath551$]).[fapp3],title="fig : " ] ( @xmath301 ) . raw data is shown in the insets , with different curves correspond to eight different values of @xmath100 collected in the interval @xmath303 $ ] ( i.e. @xmath551$]).[fapp3],title="fig : " ]
|
we analyze the spatial correlation structure of the spin density of an electron gas in the vicinity of an antiferromagnetically - coupled kondo impurity .
our analysis extends to the regime of spin - anisotropic couplings , where there are no quantitative results for spatial correlations in the literature .
we use an original and numerically exact method , based on a systematic coherent - state expansion of the ground state of the underlying spin - boson hamiltonian .
it has not yet been applied to the computation of observables that are specific to the fermionic kondo model .
we also present an important technical improvement to the method , that obviates the need to discretize modes of the fermi sea , and allows one to tackle the problem in the thermodynamic limit . as a result , one can obtain excellent spatial resolution over arbitrary length scales , for a relatively low computational cost , a feature that gives the method an advantage over popular techniques such as the numerical and density - matrix renormalization groups .
we find that the anisotropic kondo model shows rich universal scaling behavior in the spatial structure of the entanglement cloud .
first , su(2 ) spin - symmetry is dynamically restored in a finite domain in parameter space in vicinity of the isotropic line , as expected from poor man s scaling .
more surprisingly , we are able to obtain in closed analytical form a set of different , yet universal , scaling curves for strong exchange asymmetry , which are parametrized by the longitudinal exchange coupling . deep inside the cloud ,
i.e. for distances smaller than the kondo length , the correlation between the electron spin density and the impurity spin oscillates between ferromagnetic and antiferromagnetic values at the scale of the fermi wavelength , an effect that is drastically enhanced at strongly anisotropic couplings . our results also provide further numerical checks and alternative analytical approximations for the kondo overlaps that were recently computed by lukyanov , saleur , jacobsen , and vasseur [ phys .
rev .
lett . * 114 * , 080601 ( 2015 ) ] .
|
in many aspects , the existence of the irregular hydrogen - deficient ( hd ) variables remains puzzling so far .
r crb is the most known member of the post - agb group .
sakurai s object ( so , v4334 sge ) provides another , extreme case of stellar evolution .
it has been firmly established that the most abundant elements in atmospheres of r crb - like stars are helium and carbon .
determination of abundances in their atmospheres is possible only in the frame of self - consistent approach ( asplund et al .
1998 ) . still even in the case of the `` normal '' red giants that approach should be used .
otherwise , abundance determination results might be affected by significant errors ( @xmath3 0.2 - 0.3 dex , see pavlenko & yakovina 1994 for more details ) .
hd 101065 ( v816 cen ) presents another case of the peculiar stellar spectrum ( see cowley et al .
2000 ) - the strongest spectral lines in the spectrum of hd 101065 generally belong to lanthanides .
in comparatively cool ( t@xmath2 @xmath1 6500 k ) atmospheres of hd stars densities of carbon containing molecules increase ( pavlenko 2002b ) . strong molecular bands of co , cn ,
c@xmath4 appear in optical and ir spectra of sakurai s object ( pavlenko 2002b , pavlenko & geballe 2002 ) .
at photospheric levels of hotter hd stars ( t@xmath2 @xmath3 7000 k ) , the continuum opacity is governed mainly by a bound - free absorption of c atoms ( fig .
[ _ opac _ ] , see also pavlenko 1999 ) . to compute opacities due to bf absorption of c ,
n , o atoms we used topbase ( seaton 1982 ) crossections ( pavlenko & zhukovska 2002 ) . at the same time , opacity above the photosphere of hd stars with t@xmath2 @xmath1 6500 k is determined , to a large extent , by absorption of molecules contained c , n , o atoms ( pavlenko , yakovina & duerbeck 2000 ) .
naturally , this severely limits the use of odf - like methods for the computation of blanketing effects ( see discussion in mihalas 1978 ) .
we used jola and `` line by line '' models of absorption by molecules to compute opacities in atmospheres of late - type stars for kurucz ( 1993 ) , partrige and schwenke ( 1998 ) , goorvitch ( 1994 ) and harris ( 2002 ) lists .
due to drop of h@xmath6 absorption the mean opacity @xmath7 is reduced in photosheres of hd stars . as result
, they are shifted downwards higher pressure regions ( fig.[_taut _ ] ) .
we computed a grid of stellar model atmospheres of t@xmath2 = 7000 - 4000 k , log g = 0 - 1 by sam12 program ( pavlenko 2002 , 2002a ) .
sam12 is a modification of atlas12 ( kurucz 1999 ) .
opacity sampling treatment was used to account atomic and molecular absorption .
fits of theoretical seds to observed in 1997 - 1998 ones allow us to determine t@xmath2 and e@xmath8 of sakurai s object at the latest stages of its evolution ( fig .
[ _ sopt _ ] in the frame of self - consistent approach(pavlenko et al .
2000 , pavlenko & duerbeck 2001 , pavlenko & geballe 2002 ) .
fits to ir spectra allows to clearly determine an infrared excess due to emission of hot ( t @xmath3 1000 k ) dust ( fig.[_sir _ ] , see pavlenko & geballe 2002 for more details ) .
model atmospheres with t@xmath9 = 6400 - 6800 k , @xmath10 = 4.0 and abundances from cowley et al .
( 2000 ) were computed using the opacity sampling method using vald ( kupka et al .
1999 ) and dream ( dream 2002 ) line lists .
atomic lines are the main source of opacity in paczinsky s star ( fig .
[ _ prz _ ] ) .
we found , however , that a contribution of the bound - bound r - elements into opacity is rather low in comparison with other elements . the most probably , it is a consequence of incompleteness of the used line lists of r - elements .
computation of complete line lists of r - elements is of crucial importance for now .
two grids of model atmospheres of different @xmath11c/@xmath12c and c / o ratios were computed by sam12 program for opacities provided by diatomic molecules ( kurucz 1993 ) lists and hcn ( harris 2002 ) .
our model atmospheres of c - giants ( see pavlenko 2002a ) depend on input parameters , therefore fits to observed spectra are not good enough in some cases ( fig .
[ _ cg _ ] ) .
i thank soc and loc of the symposium for financial support of my participation .
i thank drs .
hilmar duerbeck , larisa yakovina & tomas geballe for very fruitful collaboration
. partial support of my investigations was provided by small research grant of aas .
|
procedure and results of computations of stellar model atmospheres and spectral energy distributions are discussed .
model atmospheres of some chemically peculiar stars are computed taking into account detailed information about their abundances : r crb - like stars of teff @xmath0 7000 k , sakurai s object ( v4334 sgr ) of 4000 @xmath1 t@xmath2 @xmath1 7000 k przybylski s star of teff @xmath0 6500 k. we show that our self - consistent approach provides a unique possibility to investigate the temporal changes of physical parameters of chemically peculiar stars . some issues of computation of model atmospheres of m and c - giants are also considered .
|
andreev spectroscopy @xcite is a powerful instrument to measure superconducting gap in a wide temperature range , up to @xmath13 @xcite .
a number of such measurements have been performed earlier with oxypnictides of the rfeaso@xmath14f@xmath15 family and with fese @xcite . here
we present systematic studies of the current - voltage characteristics ( cvcs ) and dynamic conductance @xmath16 for superconductor - constriction - superconductor ( scs ) contacts in ce-1111 , la-1111 and fese . using the intrinsic multiple andreev reflections effect ( imare ) spectroscopy , we measured the two superconducting gap values in all three fe - based materials and determined temperature dependences of the two gaps for fese .
the compounds under study belong to the class of iron - based superconductors discovered in 2008 @xcite .
these materials are characterized by a layered crystal structure ; their electron energy spectrum in the normal state contains electron and hole quasi - two - dimensional fermi surface sheets , where two superconducting condensates are supposed to be formed at @xmath17 @xcite .
to measure the superconducting gaps we used two methods : ( i ) andreev spectroscopy @xcite of single scs nanocontacts @xcite , and ( ii ) imare spectroscopy of scs - contact stacks .
the nano - sized contacts required for multiple andreev reflections spectroscopy , have been made using the `` break - junction '' technique @xcite . in this technique , breaking a bulk sample in the cryogenic environment creates a superconductor - constriction - superconductor ( scs ) junctions .
bias current flowing through the sub - mcm size constriction exceeds the superconducting critical current value and causes the contact area transition to the normal state ; as a result , the studied scs - contacts may be considered as conventional sns - junctions .
the main features of the @xmath18 curves for our scs - contacts comprise a pronounced excess current at low bias voltages and a subharmonic gap structure ( sgs ) in the @xmath19 curve .
the latter shows sharp dips at a set of bias voltages @xmath20 . for interpreting these dips we use theoretical model by kmmel _
@xcite , applicable for conductance spectra of the symmetric scs - contacts : @xmath21 as the subharmonic number @xmath22 increases , the dip amplitude decays . by plotting the @xmath23 dependence ( which must pass through the @xmath24 point ) it is easy to determine the gap value accurately . in the case of a two - band superconductor
, two distinct sgs should be observed . due to the local character of the andreev spectroscopy of scs break junctions , studies of the sgs for individual sharvin type @xcite nanocontacts
allow to gain information even in case of inhomogeneous samples . in order to observe sgs ,
the size @xmath25 of the andreev contact should be significantly smaller than the quasiparticles mean free path @xmath26 ( the ballistic regime ) @xcite . because of the layered structure of fe - based superconductors , exfoliation of the sample generates nanosteps and terraces in the @xmath27-direction and thus may form not only single scs - junctions but also arrays of the s - c - s - c- ... -s- type junctions .
the array represents a stack of several consequently connected ballistic scs - junctions causing an intrinsic multiple andreev reflections effect .
the latter is similar to the intrinsic josephson effect in sis - array @xcite .
using stacks of contacts , one can exclude surface distortion of superconductivity and observe sharp peculiarities corresponding to _ the true bulk gaps_. bias voltages for these singularities should scale with the number of contacts @xmath28 in the stack . for temperatures up to @xmath13 , the gap @xmath29 may be obtained directly by substituting to eq.(1 ) the bias voltages corresponding to the dips @xcite .
our data for three different materials are shown in figs . 1 - 5 ; they are typical for clean classical sns - contacts @xcite .
as will be shown below , the data manifest two distinct sequences of dips .
the @xmath18 and di(v)/dv - characteristics were measured by a computer controlled set - up using a 16 bit national instrument board .
the dynamic conductance spectra @xmath16 were measured by a standard modulation technique @xcite .
in this section we present andreev spectroscopy data for ce-1111 break junctions .
the results reveal the existence of two superconducting energy gaps and enable evaluating their magnitude at @xmath30k . to the best of our knowledge ,
the gap values were not measured for this material earlier .
the polycrystalline ceo@xmath0f@xmath1feas samples with @xmath31 were synthesized as described in @xcite .
-characteristics of three representative scs - contacts at @xmath32k : sample ce1 , contacts @xmath33 ( 1 ) and @xmath34 ( 2 ) , and sample ce2 , contact @xmath35 ( 3 ) .
thin line i@xmath36(v ) shows cvc for the latter contact .
the data reveal two sets of sgs corresponding to the gaps @xmath37mev ( marked with @xmath38 labels ) , and @xmath39mev ( dash - dotted vertical lines with @xmath40 labels ) .
dashed areas cover a 15% uncertainty for the large gap value .
the curves are shifted vertically , for clarity , scaledwidth=46.0% ] and @xmath41 versus @xmath42 for the studied andreev scs nanocontacts ( see fig .
1 ) : sample ce1 , contacts @xmath33 ( circles ) and @xmath34 ( triangles ) , and sample ce2 , contact @xmath35 ( rhombs ) . the averaged gap values are @xmath43mev , @xmath44mev .
* inset * shows temperature dependence of the resistance through the superconducting transition for ce-1111 sample ( @xmath2k ) , scaledwidth=42.0% ] dynamic conductance for three single scs - contacts @xmath33 , @xmath34 ( marked as 1 and 2 ) for sample ce1 , and @xmath35 ( 3 ) for sample ce2 is shown in fig .
1 , where one can see two sets of sgs . for comparison , the figure also shows the excess - current cvc for contact ( 3 ) .
the dashed areas comprise respective minima of the first set ( marked with @xmath38 labels and originating from the large gap ) and represent a 15% uncertainty . somewhat reduced intensity of the @xmath45 minima
may be caused by a slight overheating of the contact area at the highest bias voltages .
the fine structure in the bias voltage interval between @xmath45 and @xmath46 signals a large gap anisotropy of about 30% .
the small gap sgs starts with minima located at @xmath47mv ( marked with dash - dotted vertical lines and @xmath40 labels ) which have rather high relative amplitude , higher than the third andreev dip from the large gap sgs .
the sharp increase in the dip amplitude signals onset of a new sgs . beyond
the @xmath48 dips one can also see the @xmath49 dips located at @xmath50mv .
the andreev minima positions for the large and the small gap @xmath51 are plotted in fig .
2 as a function of @xmath42 .
the plot clearly demonstrates the anticipated linear dependence which proves unambiguously that the dips in fig . 1 do form two independent sgs , related with the presence of two superconducting gaps .
the slope of the two fitting lines gives @xmath52mev , and @xmath53mev , for the large and small gaps , respectively .
taking into consideration the bulk @xmath54k values ( see inset to fig .
2 ) , we find the bcs ratio @xmath55 for the large gap , and @xmath56 for the small gap .
we studied about 50 scs - andreev contacts in polycrystalline lao@xmath3f@xmath4feas ( lofa ) samples with bulk @xmath57 = @xmath58k .
the dynamic conductance @xmath16 of single contacts and nanosteps demonstrates two well - reproducible sets of sgs corresponding to the pair of independent gap values .
the number @xmath28 of elementary contacts in a stack was controlled by comparing the single contact @xmath16 spectra with those for several stacks normalized to a single junction spectrum .
figure 3 shows the @xmath16 spectra for a single contact @xmath59 ( black curve ) and for the stacks @xmath60 and @xmath61 with various number of junctions in the array ( @xmath62 , gray curve di@xmath63(v)/dv , and @xmath64 , dashed curve di@xmath65(v)/dv , respectively ) .
scaling of the sgs with properly selected number of contacts @xmath28 in nanosteps is straightforward . following the equation of kmmel _ et al . _
@xcite , we easily obtain the large gap @xmath66mev .
as for the small gap minima , peculiarities marked by arrows ( at the top of fig . 3 and in the inset )
give @xmath67mev and 1.25mev values at @xmath30k for the single contact @xmath59 and for the array @xmath60 , respectively . by tracing the @xmath9 temperature dependence
, we found the local critical temperature of the contact area @xmath68 .
the values obtained , @xmath66mev , @xmath69mev , and @xmath70k are close to the results of @xcite .
( @xmath64 junctions in a stack , dashed line ) , @xmath60 ( @xmath62 , gray lines ) , both normalized to @xmath59 ( single junction , black lines ) .
the sgs for the large gap ( @xmath38 labels ) gives @xmath66mev for all the contacts , the small gap @xmath72mev ( contact @xmath59 ) and @xmath73mev ( contact @xmath60 ) sgss are marked by black arrows and @xmath40 labels .
inset blows - up details of the small gap sgs for contacts @xmath59 and @xmath60 , scaledwidth=46.0% ] figure 4 shows normalized cvc and dynamic conductance of two - contact scs - andreev array @xmath74 in another lao(f)feas sample .
the sharp sgs related to the large gap ( marked with @xmath38 labels ) gives @xmath75mev ( see solid squares in the inset to fig .
4 ) . interestingly , all these andreev minima up to @xmath76 are double - splitted .
it seems that the doublets observed on the high - quality characteristics are caused by some anisotropy of the @xmath11 order parameter , though it has no nodes , as was shown in @xcite .
the cvc shown in fig .
4 demonstrates a small josephson supercurrent at zero bias caused by the tunneling between the sample clefts .
an sgs associated with the small gap ( @xmath40 labels and arrows ) leads to @xmath72mev value ( see open squares in the inset to fig .
the bcs - ratio @xmath77 exceeds the standard value 3.52 and thus is in favor of a strong coupling in the @xmath11 condensate . at the same time
, the small value , @xmath78 , is a result of induced superconductivity at finite temperatures in the bands with the small gap .
these values support data reported earlier in @xcite and are in close agreement with the experimental results on gdo(f)feas @xcite and our data on mgb@xmath79 @xcite
. labels ) gives @xmath75mev ( @xmath80k ) .
thin vertical lines represent the expected location of andreev minima in accordance with theoretical formula @xmath81 from @xcite .
the set of small gap peculiarities ( @xmath40 labels and black arrows ) lead to @xmath82mev .
( * inset * ) the @xmath23 dependences plotted for the sgs minima of both gaps ( using the data from the main panel ) .
lines average the experimental values , scaledwidth=44.0% ] among the new class of fe - based superconductors @xcite , layered fese has the simplest crystal structure and relatively low critical temperature @xmath13 .
polycrystalline fese samples have been grown from melt by spontaneous nucleation .
the synthesis process was described in detail in @xcite .
the intrinsic multiple andreev reflections effect was observed in fese nanosteps earlier @xcite .
spectra for scs - andreev contacts ( sample fs1 , contacts @xmath83 , @xmath60 , @xmath84 of three sns - junctions in a stack ; @xmath85k , @xmath30k ) .
two sgs at bias voltages @xmath86 corresponding to the large ( @xmath38 labels ) and the small gap ( @xmath40 labels ) yield @xmath87mev and @xmath88mev values .
, scaledwidth=46.0% ] the @xmath18 and @xmath16 characteristics for several scs - junctions formed by successive mechanical readjustments of the contact are shown in fig .
two sets of sgs with a number of dips are clearly seen .
the first set of dips ( @xmath38 labels ) gives the large gap value @xmath87mev .
the second set of dips ( @xmath40 labels ) corresponds to the small gap @xmath89mev .
it is worth noting that the dip positions and , consequently , the gap values remain unchanged under the readjustment of the contact .
this proves the high homogeneity of the sample superconducting properties in the contact area .
the superconducting gap values at @xmath30k averaged over more than 30 scs - contacts , are @xmath90mev and @xmath91mev ( @xmath6k ) .
these results agree with the preliminary data obtained with similar samples @xcite .
figure 6 shows the @xmath9 temperature dependences for two scs - contacts in fese . for the large gap ,
the @xmath92-curve lies slightly below the standard bcs - like dependence . for the small gap
, the @xmath93 dependence deviates essentially from the bcs - type curve and is in a good agreement with the calculations in @xcite . knowing the local @xmath94k
, one can calculate the bcs - ratio . for the large gap
, we obtain @xmath95 @xmath96 which exceeds the bcs value for a single - gap superconductor . on the other hand , for the small gap
, the @xmath97 ratio is much smaller than 3.52 .
such a behavior resembles the situation in mgb@xmath79 @xcite and , by parity of reasoning , can be explained by the @xmath10-space proximity effect @xcite between two superconducting condensates , where the large gap condensate plays a `` driving '' role . for fese ( sample fs1 , contact @xmath98b ; @xmath99mev ) and @xmath100 ( sample fs2 , contact @xmath101 ; @xmath102mev ) .
local @xmath103k .
single - gap bcs - like curves ( dashed lines ) , and the sample resistance @xmath104 are shown for comparison , scaledwidth=46.0% ]
it was pointed out @xcite that inelastic neutron scattering data can provide a valuable information about the symmetry of the superconducting gap in novel superconductors .
calculations showed , in particular , that a hump structure must appear in the dynamic spin susceptibility just above the @xmath105 energy in the case of an @xmath106 wave state ( the fully gapped @xmath107-wave state without sign reversal ) @xcite .
recently , the experimental linear dependence of the spin resonance energy @xmath108 on @xmath13 with the average slope @xmath109 was found for several iron based superconductors ( see , e.g. fig . 5 in @xcite ) .
within experimental errors , this dependence coincides with our plot ( fig .
7 ) of the superconducting gap @xmath110 _ versus @xmath13 for several _ iron based superconductors : ce-1111 ( present measurements ) , fese ( @xcite ) , lao(f)feas @xcite , gdo(f)feas @xcite , as well as kfe@xmath79as@xmath79 , fete@xmath14se@xmath15 , lifeas , ( see fig . 11 in @xcite and refs . therein ) .
although the scattering of data in fig . 7 is quite significant , two linear dependences emerge with @xmath111 and @xmath112 .
the coincidence of @xmath113 ( fig . 7 ) and @xmath114 ( fig . 5 in @xcite ) supports the version of a fully gapped @xmath107-wave state without sign reversal @xcite . in conclusion
, we studied properties of ceo(f)feas , lao(f)feas , and fese superconductors by scs - andreev- and imare spectroscopies .
the dynamic conductance curves for single and stack scs - contacts can not be described within the single - gap framework and evidence for the two - gap superconductivity in these compounds . for the first time
studied ceo(f)feas ( @xmath31k ) we determined the two superconducting gaps @xmath52mev , and @xmath53mev ; the respective bcs - ratios are @xmath55 , and @xmath56 . for lao(f)feas ( @xmath115k )
we also determined the two superconducting gap values @xmath116mev , @xmath117mev , leading to the bcs - ratios @xmath118 and @xmath119 , respectively ( @xmath120k ) .
we observed splitting of the sgs dips for high - quality characteristics , suggestive of an anisotropy of the @xmath11 order parameter . for fese ( @xmath121k ) our imare spectroscopy data point to @xmath90mev , @xmath91mev , and @xmath122 , @xmath123 .
the temperature dependences @xmath9 indicate the @xmath10-space proximity effect between two superconducting condensates .
the large gap bcs - ratio for all the materials studied exceeds 3.52 , indicating a strong electron - boson coupling in the `` driving '' large gap condensate .
the bcs - ratio for the small gap appears to be much less than 3.52 , thus suggesting an induced superconductivity at finite temperatures in the `` driven '' @xmath124 condensate due to a nonzero interband coupling .
finally , our data confirm a linear relation between the superconducting gap @xmath11 and magnetic resonance energy @xmath12 , valid for various fe - based superconductors .
the work was supported by russian ministry of education and sciences ( contract 11.519.11.60.12 , grant 8375 ) , rfbr ( grants 12 - 02 - 31269 , 13 - 02 - 01451 ) , dfg grants 436rus113 and for 538/bu887/4 , and dfg priority program ( spp1458 ) .
we thank t. hnke , c. hess , g. behr , r. klingeler and b. bchner for the la-1111 samples synthesis .
andreev , sov .
jetp 19 , 1228 ( 1964 ) .
r. kmmel et al . , phys . rev .
b 42 , 3992 ( 1990 ) .
a. poenicke et al . , phys .
b 65 , 220510 ( 2002 ) . g.e .
blonder et al .
b 25 , 4515 ( 1982 ) .
p. seidel , supercond .
24 , 043001 ( 2011 ) .
ponomarev , et al .
, phys . rev .
b 79 , 224517 ( 2009 ) .
shanygina et al . , jetp lett .
93 , 94 ( 2011 ) .
ponomarev et al . ,
113 , 459 ( 2011 ) .
y. kamihara et al .
130 , 3296 ( 2008 ) .
j. moreland , j.w .
ekin , j. appl . phys .
58 , 3888 ( 1985 ) .
sharvin , zh .
48 , 984 ( 1965 ) .
h. nakamura et al . , j. phys .
78 , 123712 ( 2009 ) .
ponomarev , a.v .
rakhmanina , prib .
eksp . 5 , 120 ( 1970 ) .
khlybov et al . , jetp lett . 90 ( 2009 )
. m. yashima , et al .
, j. phys .
78 , 103702 ( 2009 ) .
ponomarev et al . , solid st .
129 , 85 ( 2004 ) .
r. khasanov et al .
104 , 087004 ( 2010 ) .
ponomarev et al . , jetp lett .
79 , 484 ( 2004 ) .
kuzmichev et al .
, solid st . comm .
152 , 119 ( 2012 ) .
nicol , j.p .
carbotte , phys . rev .
b 71 , 054501 ( 2005 ) .
golubov et al .
, phys . rev .
b 51 , 1073 ( 1995 ) .
yanson et al . , phys . rev .
b 67 , 024517 ( 2003 ) .
s. onari et al . , phys .
b , 81 , 060504 ( 2010 ) .
shin - ichi shamoto et al .
, phys . rev .
b 82 , 172508 ( 2010 ) .
|
using the `` break - junction '' technique we prepared and studied superconductor - constriction - superconductor nanocontacts in polycrystalline samples of fe - based superconductors ceo@xmath0f@xmath1feas ( ce-1111 ; @xmath2k ) , lao@xmath3f@xmath4feas ( la-1111 ; @xmath5k ) , and fese ( @xmath6k ) .
we detected two subharmonic gap structures related with multiple andreev reflections , indicating the presence of two superconducting gaps with the bcs - ratios @xmath7 and @xmath8 , respectively .
temperature dependences of the two gaps @xmath9 in fese indicate a @xmath10-space proximity effect between two superconducting condensates . for the studied iron - based superconductors we found a linear relation between the gap @xmath11 and magnetic resonance energy , @xmath12 .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore
|
the interstellar medium ( ism ) in galaxies is excited , dissociated , and ionized by far - ultraviolet ( fuv ) photons produced by young o and b stars .
atomic gas in the ism recombines into molecular form mainly through the catalytic action of dust grain surfaces . in particular , hydrogen nuclei in the ism cycle
repeatedly from the molecular ( ) to the atomic ( ) phase and back again , at rates depending on the incident fuv flux @xmath3 , the total volume density @xmath2 of the gas , and the dust - to - gas ratio @xmath8 .
regions in the ism where the physics is dominated by fuv photons are called photodissociation regions ( pdrs ) .
the surfaces of giant molecular clouds ( gmcs ) are important ( and ubiquitous ) examples of pdrs in galaxies .
the physics of pdrs has been explored in detail over the past @xmath9 years by many workers including d. j. hollenbach , b. t. draine , a. dalgarno , j. h. black , a. g. g. m. tielens , e. van dishoeck , j. le bourlot , and their students and collaborators .
one major focus of this work has been to explain the @xmath10% line - to - continuum ratios of the far - infrared lines of ] and in galactic sources observed from high - altitude aircraft and balloon platforms . with the advent of more sensitive space - based observations ,
the models were extended to include the weak rotational - vibrational spectrum of excited observed on the active pdr surfaces of gmcs that are exposed to relatively intense fuv fluxes from nearby young stars .
an excellent review of the observational and theoretical state of the field is given by @xcite .
pdr model computations have become very detailed and comprehensive , raising the possibility that observations of multiple spectral lines including those from trace molecules such as co can be used to `` invert '' the models in order to determine physical conditions in the ism .
for example , a set of such models has been computed by @xcite for use in the interpretation of far - infrared and submillimeter spectra from galactic and extragalactic sources .
however , we wish to point out that , although there is qualitative agreement on most of the physics , there are differences in the details of the various model codes currently in use around the world as well as differences in the numerical values of the parameters adopted . these differences can lead to disagreements in the values derived for physical conditions even for the same observational data : an example of such differences is illustrated in @xcite .
efforts are underway to intercompare models from different groups in a more systematic manner ( e. van dishoeck 2003 , private communication ) .
the results obtained from any one model computation ( such as the one we use here ) can therefore be revealing in general , but the specific numerical values obtained should be taken with some caution .
the purpose of the present paper is to draw attention to the fact that the produced from photodissociated can also be a useful diagnostic for determining physical conditions in pdrs . in particular
, the combination of the 21 cm and 2.6 mm co(10 ) lines provides a means of independently estimating @xmath2 and @xmath3 in distant pdrs using ground - based radio astronomy data .
these two radio lines form a diagnostic that is particularly useful in the area of parameter space where @xmath3 @xmath4 , 10 @xmath11 ; this range of fuv flux and total density is representative of most of the volume of the ism in galaxies .
our point of departure is the paper by @xcite that describes the results of steady state computations on a specific , extensive set of pdr models over a wide range of physical conditions .
these models are simple one - dimensional semi - infinite slabs of gas with constant density @xmath12 of h nuclei subjected to an equivalent one dimensional flux of fuv photons @xmath3 measured in units of ergs ps . over the photon energy range 613.6 ev .
sr is @xmath3 @xmath13 ( @xcite , see our appendix [ app : isrf ] and draine s fig .
the line brightness is calculated along a ray normal to the surface of the slab .
see @xcite for further details of their models .
the incident fuv flux dissociates and heats the cloud surface . at some depth into the cloud
, the dissociation rate has decreased enough to allow co molecules to survive , and the density and heating rate are still high enough to excite rotational line emission from these molecules . among many important results ,
@xcite have computed the 2.6 mm co(10 ) brightness , @xmath14 , emanating from the cloud surface ; this is the most widely studied molecular line in the ism . in figure
[ fig : modelcohi]a we present their results for @xmath15 as a contour diagram of constant surface brightness ( see their fig .
11 ) ; we have extended the range of @xmath2 and @xmath3 down to 1 and 0.03 , respectively , and labeled the contours with their original `` theoretical '' units of ergs ps . , as well as with `` co observer '' units of k .
this latter conversion is made according to : @xmath16 where @xmath17 is the measured emergent rayleigh - jeans brightness temperature assuming the cloud is completely resolved , @xmath18 is the profile velocity width in , @xmath19 mm , and @xmath20 is boltzmann s constant .
we note that figure [ fig : modelcohi ] assumes that the dust - to - gas ratio @xmath8 is that of the local ism near the sun , and only one side of the slab contributes to the observed emission .
a further caution concerns the value of the density @xmath2 inferred from a specific observed value of the co(10 ) intensity in figure [ fig : modelcohi]a .
there is a general trend in most pdr models to underestimate the co abundance , so that the density inferred from a given observed value of the co(10 ) line is often higher than that obtained from other independent diagnostics .
this is another example in which comparisons among the different modeling efforts will be very useful .
the co(10 ) surface brightness in figure [ fig : modelcohi]a is governed by a complex interplay of processes involving the dissociation and reformation of co molecules , heating and cooling in the ism , and the detailed excitation and macroscopic radiative transfer of the co(10 ) line . by contrast , the 21 cm line intensity is somewhat easier to compute , since in general this line is optically thin and independent of the kinetic temperature of the ism ; we can therefore ignore the physics of the 21 cm line formation and simply compute the amount of produced .
incident fuv photons in the wavelength range 9121108 ( 13.611.2 ev ) also dissociate the on the surface of the gmc .
@xmath21 @xcite .
] this process was first described in the literature by @xcite and is by now well known : it has been summarized by @xcite , who also provide several other references .
the dissociated creates a `` blanket '' of atomic gas on the surface of the gmc ; the atoms recombine into on dust grains and are then returned to the gaseous ism .
approximate analytic expressions for the steady state column density @xmath22 of the atomic gas have been derived by several authors under a variety of limiting assumptions ( e.g.@xcite ) and are discussed in more detail in appendix [ app : modelhiboth ] .
computations of @xmath22 in the context of the present `` standard model '' were first presented in @xcite ( see their fig .
an updated calculation using the latest code and the model parameters in @xcite is presented as the dotted curves in figure [ fig : modelhiboth ] .
the results differ slightly from those of @xcite , owing to small differences in the assumed values of dust abundance , atomic abundances , and a different formation rate .
this leads to changes in the depth of the
transition , and hence to different values of @xmath22 ; the differences are , however , generally less than about a factor of 2 .
a problem arises in attempting to combine the computational results for co(10 ) with those for .
the `` standard model '' used here is bathed in a uniform flux of cosmic rays producing a constant ionization rate of ps . .
these cosmic rays also produce a faint extended distribution of atoms owing mainly to ion chemical reactions everywhere in the cloud .
however , as we see below , such a faint extended distribution of will in general be removed from the observations by the data reduction process , so we would prefer to use models for @xmath22 _ without _ a cosmic - ray component .
nevertheless , cosmic - ray ionization is important for the chemistry in the model , thereby controlling to some extent the abundance of co molecules , so we ought to keep it in the computations of the co(10 ) line . as a compromise we have chosen to use the approximate analytic expression for @xmath22 since it does not include any effects of cosmic rays .
this approximation is presented in appendix [ app : modelhiboth ] where it is compared with the standard computational model : it is an excellent fit to the model over the full range of parameter space important for this paper and provides a better context in which to compare the theory with the observations . the adopted result for the column
is shown in figure [ fig : modelcohi]b .
the conversion from column density in figure [ fig : modelcohi]b to 21 cm emission intensity in k is : @xmath23 for @xmath22 in units of atoms , and assuming the optical depth of the atomic gas to the 21 cm line radiation is small . a noteworthy feature of figure [ fig : modelcohi]b is _ the constancy of n ( ) for a constant ratio of @xmath3/@xmath2 _ ( see also eq.[eqn : dissociate1 ] ) ; the same column density can be produced in high fuv flux , high - density environments or in low fuv flux , low - density environments .
in addition , _ the column decreases with increasing density at a given value of fuv flux _ : this occurs because the destruction rate for varies as @xmath2 , but the formation rate varies as @xmath24 ( since the grain density is proportional to @xmath2 ) .
figures [ fig : modelcohi]a and [ fig : modelcohi]b show that the contours of constant co(10 ) emission @xmath15 and constant column density @xmath22 behave in a complementary fashion .
this suggests that over much of the @xmath2@xmath3 plane an observational determination of @xmath15 and @xmath22 for the same pdr would permit independent estimates to be made of both @xmath2 and @xmath3 . in figure
[ fig : modelboth ] we show the two model computations from figures [ fig : modelcohi]a and b superposed , along with several additional contours and a `` box , '' which will be discussed further below .
the models become progressively less reliable in the top left corner of the plots for both @xmath15 and @xmath22 .
this occurs because the assumption of a steady state model becomes increasingly unreliable in the high @xmath3
low @xmath2 section of the diagram ; in this region , radiation pressure causes grains to drift with speeds of the same order as the gas turbulence speeds ( see @xcite , fig . 2 ) .
furthermore , this is also the area of the diagram where large column densities are predicted ; however , values of @xmath22 in excess of a few @xmath25 @xmath26 are not likely to be observed since the 21 cm line may become optically thick in this regime : @xmath27 for spin temperatures of @xmath28 k , profile fwhms of @xmath29 , and @xmath30 , values that probably represent the extremes typical for the observations .
it should be emphasized that the present calculations of surface brightness can be applied directly only to observations for which the sources of emission are resolved . in any realistic situation ,
the emission probably arises from several pdr surfaces . in this case
, the density and fuv intensity derived from our model using observations of @xmath15 and @xmath22 represent ensemble averages of the distribution of values for @xmath2 and @xmath3 .
@xcite have shown that these averages are somewhat biased toward higher fuv fields ; bias toward high or low density depends on the cloud distribution .
the x axis of figure [ fig : modelboth ] refers to the total volume density in our computational model of an isopycnic cloud . in a real gmc
this is likely to be close to the actual density at the depth in the cloud where the co(10 ) line is formed .
however , it is not likely to be representative of the density of dissociated . this is because the various heating processes at the surface of the gmc ( including photodissociation ) will lead to gradients in temperature and , therefore , if the pressure is roughly constant , to gradients in density .
for example , the model computations of @xcite for relatively low fuv fluxes used by allen et al . to denote the fuv flux has very nearly the same value as we use here , @xmath31
@xmath3/0.85 for irradiation over @xmath32 sr .
see appendix [ app : isrf ] for a discussion of the various definitions of fuv values used in the literature on pdrs .
] of @xmath3 @xmath33 show that the density at the co(10)-emitting layer is about 5 times that at the interface .
their figure 4 shows that the co emission emanates from a depth such that @xmath34 , where the temperature is just below 5 k ( see their fig .
3b ) , whereas their figure 2 shows the
interface is effectively at very small values of @xmath35 , where the temperature is about 25 k. we have calculated temperatures for higher fuv flux levels @xmath3 = 3 : the surface temperatures are higher in this case , @xmath36 k at the interface , and @xmath37 k at the co emission layer , for a ratio of @xmath38
. this ratio will continue to rise for yet higher photon fluxes , since the increased surface heating will result in further increases in surface temperature , but the increase in temperature deeper in the clouds where the co(10 ) emission is formed will be less rapid because of more effective cooling .
if the density and temperature are linked by an isobaric condition , the density observed in the bulk of the may be a factor of 510 lower than that inferred for the co(10 ) line .
it is clear from this discussion that constant - density models for pdrs such as the one we have assumed here are not likely to be very realistic .
more sophisticated models , for example assuming pressure equilibrium , or a full computation with various forms of micro- and macro - turbulent pressure ( e.g. , @xcite ) over the range of the @xmath3@xmath2 parameter space may provide more realistic results and ought to be explored .
figure [ fig : modelboth ] clearly shows the complementarity of 21 cm and 2.6 mm co(10 ) radio line emission as diagnostics of pdrs . over the range of validity of our calculations , the co(10 ) line brightness @xmath15 depends mostly on the volume density of the gas for @xmath39 , no matter how intense the fuv flux . to the fuv flux
comes about because increasing @xmath3 merely leads to more photodissociation of the surface co , and the co - emitting layer retreats deeper into the static cloud where the temperatures are about the same as they were for the case of lower @xmath3 . ] on the other hand , for constant @xmath2 the brightness increases nearly linearly with increasing fuv fluxes up to @xmath3@xmath40 , becoming logarithmic for higher fuv fluxes .
accordingly , the combination of measurements of @xmath22 and @xmath15 from the same pdr can provide unique values for @xmath2 and @xmath3 over a substantial range in the diagram .
how can extraneous emission from gas that lies far outside the cloud be identified ?
the spatially layered structure of pdrs provides a direct clue , as has been convincingly shown e.g. by the observations of the orion bar region ( see fig . 2 in @xcite ) on the @xmath41 pc scale . a more or less edge - on viewing angle
is required for a certain identification , and one ought to remove any extended emission that is associated with other clouds along the line of sight or produced by other excitation mechanisms .
galaxies that are viewed at low to intermediate inclination angles are favorably oriented for this identification to work , and @xcite and @xcite have shown that the signature morphology can be found at 100 pc scales in the highest resolution images of nearby galaxies m81 and m101 .
as to the co , it is generally assumed that this arises in pdrs , although the morphological signature is not always clear .
figure [ fig : modelboth ] provides a method for determining @xmath2 and @xmath3for a specific pdr if @xmath22 and @xmath15 can be determined observationally . as an example
, we consider the galactic object g216
-2.5 ( also called `` maddalena s cloud '' ) which has been proposed to be a large pdr in the galaxy by @xcite . determining the column density of that part of the associated with g216
-2.5 is made quite difficult by confusion from unassociated superposed along the line of sight through the galaxy .
several estimates can be made from the single - dish observations reported by @xcite .
first , and perhaps simplest , we estimate that fraction of the total averaged profile shown in their figure 1 that is in the velocity range of the co(10 ) emission ( also shown on the same figure ) ; the profile shows three overlapping peaks and integrates to @xmath42 k @xmath43 @xmath44 k = .
the averaged @xmath15 profile corresponds to the central peak and is @xmath45 wide @xcite , so the average amount of column associated with the co emission is , by this estimate , probably not more than @xmath46 = . in their 3.2.2 ,
@xcite carry out a more involved analysis with area integrations of the in the velocity range 1638 and conclude that the `` excess '' associated with the pdr is @xmath47 ; the uncertainty in this method is approximately a factor of 2 ( r. maddalena 2002 , private communication ) . a value in the range
may therefore be considered typical for this cloud , with some preference for the lower end of that range . from figure 1 of @xcite , the total @xmath15 profile over the whole cloud integrates to @xmath48 k .
the co(10 ) map for the cloud integrated over the same velocity range of 1638 as the is shown in figures 2 and 4 of @xcite , based on earlier data of @xcite .
maps in @xcite are similar in shape but brighter by about a factor of 2 when compared to the earlier map of @xmath15 in fig . 2 of @xcite .
this is not likely to be an effect of differences in the velocity range over which the data have been integrated , since that range is smaller than the range of 1540 used by maddalena & thaddeus .
we use the more recent results of williams & maddalena . ] from figure 2 of @xcite the co(10 ) brightness ranges from 37 contours with a few positions reaching 10 contours . a range of 38 contours or 616
k therefore seems representative of the co data , again with some preference for the lower end of that range .
the range of values for @xmath15 and @xmath22 obtained above is plotted as a black box on our figure [ fig : modelboth ] . in the context of our model , these values describe gas with a density range of and an incident fuv flux of 0.17 of the standard value @xmath3near the sun , with likely values being approximately and 0.8 , respectively .
although no independent measurements of the total volume density are available for g216 -2.5 , the value we have obtained is in the range generally indicated for gmcs in the galaxy ( @xmath49 , e.g. , @xcite ) . as to the fuv flux , @xcite estimated @xmath3 @xmath10 from what is known about the two nearby young stars identified in their study as likely to be responsible for the photodissociation .
our model thus agrees well with the observations .
the practical detection limits for the 21 cm line are @xmath50 and @xmath51 k for single - dish survey and interferometer array imaging observations , respectively , corresponding to @xmath52 and .
these are shown as the dashed ( single - dish survey ) and dotted ( interferometer array ) lines for @xmath22 on figure [ fig : modelboth ] .
the corresponding values for the 2.6 mm co line are @xmath10 k ( dashed line : single dish survey ) and @xmath53 k ( dotted line : interferometer array ) . in all cases , this assumes that the gas clouds are resolved ; otherwise these limits need to be increased further by the ratio of the beam area to the cloud area . locating the detection limits on our figure [ fig : modelboth ]
, we can conclude that the range in parameter space over which the combination of and co(10 ) can provide useful constraints on @xmath2 and @xmath3 is given approximately by @xmath3 @xmath4 , 10 @xmath54 .
these boundaries are set by the decreasing sensitivity of the emission at high fuv flux levels ( @xmath22 increases only logarithmically for high @xmath3 at constant @xmath2 ) and by observational detection limits ( @xmath15 disappears at low @xmath2 owing to insufficient excitation of the transition ; disappears at high @xmath2 as the gas stays mostly molecular ) .
many observers who report synthesis imaging ( interferometer array ) data in the literature use units of jy beam@xmath55 instead of k. the required conversion factors are ( unfortunately ) telescope dependent ; appendix [ app : conversions ] provides an approximate recipe for these conversions .
we have shown that a combination of observations in the 21 cm line of and the 2.6 mm line of co can provide a useful diagnostic for physical conditions in distant pdrs . both of these lines
are readily observed in the galaxy and in nearby galaxies using ground - based single - dish and interferometer - array radio telescopes . the useful range in parameter space for this combined radio line diagnostic is approximately bounded by @xmath3 @xmath4 , 10 @xmath5 , a range that covers a large fraction of the volume of the ism in galaxies .
the unique layered morphology of the emission in pdrs provides a means of identifying that part of the and co emission that is related when that morphology is observable , but confusion along the line of sight makes this separation difficult in the galaxy .
this method will be especially useful for the interpretation of high - resolution and co synthesis imaging data in nearby galaxies where the pdr morphology can be more easily identified .
improvements in the model parameters and in the details of the computations are anticipated in the near future that will permit even more precise values of @xmath3 and @xmath2 to be obtained .
we are grateful to bruce draine and david hollenbach for discussions on the physics of pdrs .
david also provided very helpful comments on earlier drafts of this paper .
the referee offered many useful suggestions that have also been incorporated .
the column density in a pdr is calculated with the same physics used to determine the excitation of the near - infrared fluorescence lines . a logarithmic form for the analytic solution to the equation for formation destruction equilibrium was first given by @xcite ( see also eq . a2 in @xcite . ] and eq . 3 in @xcite ) .
the model is a simple semi - infinite slab geometry in statistical equilibrium with fuv radiation incident on one side .
the solution gives the steady state column density along a line of sight perpendicular to the face of the slab as a function of @xmath56 , the incident uv intensity scaling factor ( see appendix [ app : isrf ] ) , and the total volume density @xmath2 of h nuclei .
sternberg s result is : @xmath57 } , \label{eqn : nhi}\ ] ] where ; [ cols=">,^,<,>,^ , < " , ] equation [ eqn : nhi ] has been developed using a simplified three - level model for the excitation of the molecule and is applicable for low - density ( @xmath58 ) , cold ( t @xmath59 k ) , isothermal , and static conditions , and neglects contributions to @xmath22 from ion chemistry and direct dissociation by cosmic rays .
the quantity @xmath60 here ( not to be confused with @xmath3 used previously ) is a dimensionless function of the effective grain absorption cross section @xmath61 , the absorption self shielding function @xmath62 , and the column density of molecular hydrogen @xmath63 : @xmath64 the function @xmath60 becomes constant for large values of @xmath63 due to self shielding @xcite . using the parameter values in this equation adopted by @xcite , we have : @xmath65\ ] ] where @xmath2 is in .
this is a steady state model , with continually forming from on dust grain surfaces , and continually forming from by photodissociation . in order to compare this result with the `` standard model '' computations we need to relate @xmath56 to the @xmath3 used by @xcite ;
this is because @xcite and @xcite use different normalisations for the fuv flux ( see appendix [ app : isrf ] ) .
when distributed sources illuminate an fuv - opaque pdr over @xmath32 sr , the conversion is @xmath66 ( see footnote 7 in @xcite ) , resulting in @xmath67 .
we have fitted the analytic expression for @xmath22 to the model computations in the range in which cosmic ray dissociation is not a major contributor , roughly for @xmath3 @xmath68 , @xmath69 .
the result is that no consistent improvement is obtained by using any value for the coefficient of @xmath3 other than the value 106 deduced above , although a modest improvement is obtained by using a slightly larger value for the leading coefficient in the equation , , corresponding to a value of @xmath70 for the effective grain absorption cross section . with these small adjustments ,
our final equation is : @xmath71 { \rm ~cm}^{-2 } , \label{eqn : dissociate1}\ ] ] resulting in an @xmath72 value for the fit that is everywhere in excess of 99% over the fitted range of @xmath2 and @xmath3 . in figure
[ fig : modelhiboth ] we show values from equation [ eqn : dissociate1 ] plotted as solid lines together with dotted contour lines from our standard numerical model .
the agreement is generally good over much of the @xmath2@xmath3 parameter space of interest here ; differences occur mainly in the top left corner of the diagram , and at low values of fuv flux . in the top left corner the analytic formula under - predicts the amount of column density computed from the standard model by about 30% owing to production by ion chemistry reactions such as @xmath73 , @xmath74 , and @xmath75 ( where pah is a polycyclic aromatic hydrocarbon ) , which are important at high @xmath3 and low @xmath2 . at values of @xmath3 @xmath76 the contours of @xmath22 for the numerical model become vertical ; this is because the standard model includes a low level of cosmic ray ionization ( see main text ) which contributes a small amount of even for @xmath3 = 0 .
@xcite characterized the emission from photodissociation regions ( pdrs ) in terms of the `` strength '' of the fuv radiation field , @xmath3 , illuminating the medium at energies between 6 and 13.6 ev .
numerical values for @xmath3were presented in units of the `` habing field , '' following earlier pdr modeling by @xcite , who expressed @xmath3 in units of the `` equivalent @xcite flux of ergs @xmath26 s@xmath55 appropriate to the average interstellar medium . ''
however , other characterizations of the fuv background also appear in the literature for the `` typical '' and local ism .
@xcite has developed an analytic fit to many of these , which he defined as the `` standard uv background '' from 513.6 ev .
although numerical values for the @xcite and @xcite representations of the average interstellar background can be interconverted , draine s figure 3 makes it evident that they have different spectral shapes .
furthermore , the `` strength '' of the interstellar fuv field is described in the literature by several different quantities ( e.g. , energy density , photon intensity , energy flux ) using a variety of units .
some authors use a single symbol to report normalized values of their field descriptor , while others specifically represent the normalized quantity by a ratio .
accordingly , the purpose of this appendix is to summarize the approach taken by habing and draine in developing their descriptions of the isrf at ultraviolet wavelengths and to interrelate the representations of that field adopted by several authors whose work is germane to our paper .
@xcite calculated radiation energy density values , @xmath77 , at 1000 , 1400 , and 2200 which were averaged `` over a considerable area '' in space , using bandwidths of 9121040 , 13001450 , and 19002400 , respectively .
his recommended values for the averaged energy density @xmath78 were @xmath79 however , habing s results depend strongly on assumptions made concerning the wavelength dependence of interstellar extinction and the presence of diffuse galactic light .
@xcite noted that these are `` two areas for which there existed little or no observational data at the time of his work . ''
@xcite converted these and other existing calculations of the interstellar uv energy density into a quantity that he defined as the angle - averaged photon flux @xmath80 where @xmath81 is the energy , @xmath82 represents planck s constant , and @xmath83 is the speed of light .
the resulting values and a few observational data points are plotted on his figure 3 , together with an analytical expression that is `` in good agreement with all of the above results over the range 513.6 ev . ''
that expression , @xmath84 where @xmath81 is measured in ev , was referred to by draine as the `` standard uv background '' in the ism . since it is computed on a `` per steradian '' basis , this quantity is actually an angle - averaged photon intensity .
we have used equation [ eqn : photflux ] to convert the @xcite energy density values to angle - averaged photon intensities ( denoted @xmath85 ) at the three wavelengths reported in his paper , and they agree well with the corresponding points on figure 3 of @xcite .
we have also computed values for draine s `` standard '' uv background at the same wavelengths , using equation [ eqn : fitphotflux ] . at 1000 , @xmath86 @xmath87 , @xmath88 ( in the same units ) , and @xmath89 .
thus , the photon intensity computed from draine s fit at this wavelength is 1.710 times that found by converting habing s calculated energy density to these units .
this result is identical to the value computed by @xcite at 1000 , and ( to one decimal point ) to the value given in footnote 7 of @xcite .
the same result could also be obtained by first converting draine s photon intensity to an energy density for direct comparison to habing s ( 1968 ) calculation . @xcite
chose habing s result at 1000 as the basis for normalization because in `` neutral regions , pumping of is primarily effected by far - ultraviolet photons in the 1110912 range '' .
the energy flux over @xmath90 sr at 1000 is @xmath91 ergs @xmath26 s@xmath92 , where @xmath93 is the mean intensity .
the values for these quantities at 1400 are @xmath86 , @xmath94 , and @xmath95 . in this case ,
draine s fit to the angle - averaged photon intensity is 1.965 times the converted habing result , and the energy flux is @xmath96 . at 2200 , @xmath86 , @xmath94 , and @xmath97 .
the `` standard '' draine result is 1.211 times that calculated by habing at this wavelength , and the energy flux @xmath91 .
draine s standard curve can be integrated from 6 to 13.6 ev to find the total photon intensity , i.e. , photons @xmath26 s@xmath55 sr@xmath55 .
when this result is multiplied by the average energy per photon ( i.e. , ergs ) over that interval , the draine flux over @xmath90 sr is @xmath98 ergs @xmath26 s@xmath55 .
if we assume that the photon intensity ratio found earlier in the 11.913.6 ev band , e.g. , @xmath99 , applies over the entire fuv spectrum ( i.e. , 613.6 ev ) , then @xmath100 ergs @xmath26 s@xmath55 .
this is the equivalent habing flux . since both ultraviolet
normalization procedures are still used in the literature , confusion can arise when comparing descriptors of the isrf .
@xcite represent that field by an `` incident fuv flux '' , @xmath3 , that is already normalized by the equivalent habing flux of the average ism ( see their table 1 ) .
normalization to the habing field was also used by @xcite and @xcite .
the use of this normalization in these models is purely historical , having been incorporated into early versions of the underlying code in support of pdr modeling begun in the 1980 s ( m. wolfire 1999 , private communication ) .
in contrast , @xcite and @xcite compute the `` uv photon intensity '' ( i.e. , @xmath101 ) by introducing a `` uv intensity scaling factor '' @xmath56 which is defined to be unity for draine s fit to the average interstellar field intensity @xmath102 . in @xcite and @xcite ,
the @xcite standard result is re - expressed as @xmath103 where @xmath104 is expressed in nanometers . when sternberg s representation of draine s standard background ( i.e. , eq .
[ eqn : drafreq ] ) is integrated from 6 to 13.6 ev , and the resulting value , photons @xmath26 s@xmath55 , is multiplied by the average energy per photon , the total fuv flux over @xmath90 sr is @xmath105 ergs @xmath26 s@xmath55 . once again , we find that @xmath106 provided the ratio @xmath89 found earlier between 11.9 and 13.6 ev ( i.e. , 9121040 ) is assumed to apply over the entire fuv band
. equation [ eqn : drafreq ] can be obtained from equation [ eqn : fitphotflux ] by multiplying the latter by @xmath107 sr and ev hz@xmath55 , provided @xmath81 is replaced by @xmath108 ev .
thus , @xmath109 we note that equation [ eqn : drafreq ] represents a photon flux rather than the photon intensity , since it has been averaged over solid angle .
@xcite corrected this problem in their equation a1 by multiplying equation [ eqn : drafreq ] by @xmath110 , resulting in corrected units of photons @xmath26 s@xmath55 hz@xmath55 sr@xmath55 .
when this correction is made , equation [ eqn : drafreq2 ] becomes @xmath111 .
@xcite explicitly introduced field normalization through the ratio @xmath112 , where @xmath113 is defined as the uv field intensity , although they also called it an fuv energy flux .
unfortunately , the subscript `` uv '' is dropped later in their paper where the ratio is shown as @xmath114 , increasing the potential for equating the numerator ( incorrectly ) with sternberg s parameter @xmath56 upon casual inspection .
the divisor , @xmath115 ergs @xmath26 s@xmath55 sr@xmath55 , was determined from the `` uv field '' results in the solar neighborhood presented by @xcite .
although not explicitly stated when @xmath116 was defined , their introduction identifies the band of interest as extending from 912 - 2000 (i.e .
, 6.2 - 13.6 ev ) . over @xmath107 sr , the flux corresponding to the madden et al .
value of @xmath116 is @xmath117 ergs @xmath26 s@xmath55 , and @xmath100 if we continue to assume that @xmath89 .
the key issue for this paper is how to relate the dimensionless field descriptor chosen by @xcite , i.e. , @xmath3 , to the dimensionless quantity used by @xcite , @xmath56 , or by @xcite , @xmath112 .
since @xmath56 was normalized by draine s standard background between 6 and 13.6 ev , and @xmath116 was computed by integrating the same curve over a similar energy range ( the lower limits may differ by 0.2 ev ) , @xmath112 is equivalent to @xmath56 .
we illustrate this circumstance by comparing the flux computed earlier from sternberg s representation of draine s standard curve ( i.e. , from eq .
[ eqn : drafreq ] ) , given by @xmath118 ergs @xmath26 s@xmath55 , with the flux computed ( using fewer significant figures ) from the value of @xmath116 used by @xcite .
the latter is @xmath117 ergs @xmath26 s@xmath55 . this conclusion is not affected by whether or not the @xcite expression is multiplied by a solid angle factor ( i.e. , @xmath110 ) provided his numerator is multiplied by the same factor .
the dimensionless parameter @xmath3 @xmath119 / [ 4\pi \int f_h(e ) de]$ ] is a flux ratio that is computed over the same spectral band as @xmath116 , where the denominator @xmath120 in the highest energy band ( i.e. , 11.913.6 ev ) addressed by @xcite .
accordingly , over that band @xmath3 @xmath121 . since @xmath116 ( with units of ergs @xmath26 s@xmath55 sr@xmath55 ) represents an intensity that is referenced to draine s standard curve , it can be written as @xmath122 , whereby @xmath3@xmath123 . inserting this into equation 12 of @xcite , the argument of the logarithm , i.e. , @xmath124 + 1 $ ] can be written as @xmath125 + 1 $ ] , in agreement with equation [ eqn : dissociate1 ] when the latter is written to accommodate irradiation over @xmath107 sr .
it is common practice to report results from imaging synthesis observations of radio spectral lines as contour diagrams of constant surface brightness using `` j beam@xmath55 '' as the unit . ''
( e.g.fig . 4 in @xcite ) .
] this unfortunate combination of units is telescope dependent through the use of the beam unit in the denominator .
the conversion to telescope - independent units such as k ( which we use here in eqs .
[ eqn : tmbco ] and [ eqn : tmbhi ] ) requires knowledge of the beam area of the radio telescope in steradians , which is not well defined for any radio synthesis imaging telescope owing to the absence of information at short interferometer spacings .
fortunately ( at least for now ) , most data processing systems perform an image restoration step that sets the final point spread function to be a two - dimensional gaussian .
this provides a definite ( but not necessarily accurate ) value for the two - dimensional integral over the beam , and thereby `` solves '' the unit conversion problem .
if we take the fwhms of this gaussian to be @xmath126 and @xmath127 , then the conversion for 21 cm image contours may be done with the relation with the same units as the previous equation . we emphasize again that these temperatures are well defined only for completely resolved sources .
if the telescope beam does not resolve the source , the data can provide only a lower limit to the true brightness ; such lower limits are usually called `` main beam brightness temperatures '' and labelled @xmath132 .
|
the gas at the surfaces of molecular clouds in galaxies is heated and dissociated by photons from young stars both near and far .
resulting from the dissociation of molecular hydrogen emits hyperfine line emission at @xmath0 cm , and warmed co emits dipole rotational lines such as the @xmath1 mm line of co(10 ) .
we use previously developed models for photodissociation regions ( pdrs ) to compute the intensities of these and co(10 ) lines as a function of the total volume density @xmath2 in the cloud and the far ultraviolet flux @xmath3 incident upon it and present the results in units familiar to observers .
the intensities of these two lines behave differently with changing physical conditions in the pdr , and , taken together , the two lines can provide a ground based radio astronomy diagnostic for determining @xmath2 and @xmath3 separately in distant molecular clouds .
this diagnostic is particularly useful in the range @xmath3 @xmath4 , 10 @xmath5 , which applies to a large fraction of the volume of the interstellar medium in galaxies . if the molecular cloud is located near discrete sources of far uv ( fuv ) emission , the pdr
generated and co(10 ) emission on the cloud surface can be more easily identified , appearing as layered `` blankets '' or `` blisters '' on the side of the cloud nearest to the fuv source .
as an illustration , we consider the galactic object g216 -2.5 , i.e. `` maddalena s cloud '' , which has been previously identified as a large pdr in the galaxy .
we determine that this cloud has @xmath6 and @xmath3 @xmath7 , consistent with other data .
|
our main contribution in this paper is to further investigate the model theory of non - algebraic generic structures . in a language @xmath0 with a ternary relation , for each @xmath5 $ ] one can associate to the class of finite @xmath0-structures some certain function @xmath4 called a _ predimension _ ( ) . in the present work we mainly focus on understanding particular _ generic _ structures ( ) , @xmath6 which were introduced by hrushovski in @xcite and were later studied by various authors in @xcite , , and . for a rational @xmath8
it has been proved by evans and wong that all finite graphs are interpretable in @xmath9 and as a consequence , @xmath10 has the strict order property and is undecidable .
later , brody and laskowski gave another source of undecidability for @xmath11 by interpreting robinson arithmetic in a subtheory of @xmath12 these results suggest that @xmath10 should be assumed to be quite wild " .
however , the starting point for investigating theory of generic structures is to determine whether the type of tuples can be fully determined in terms of their corresponding closures .
this happens in particular within the theory of tame " generic structures , where the type of a tuple is given completely by ( boolean combinations of ) certain @xmath13-formulas which indicate the existence of certain diagrams in the closure of that specific tuple .
our first result suggests that , despite the fact that the theory of @xmath11 behaves in a wild way , it still shows some levels of tameness by permitting the type of tuples to be fully described by their closures ; although this description is obtained through more complex formulas , called _
closure formulas_. in addition , this result provides the means to further distinguish the theory of a certain ultraproduct of generic structures , hence strengthening the result by brody and laskowski ( theorem 4.9 of ) .
more on the wildness side , we further refine the techniques developed in and show that the structure @xmath14 is interpretable in @xmath15 hence we prove that @xmath10 does not have the finite model property .
this answers a question posed by evans and wong in .
[ [ setting . ] ] setting .
+ + + + + + + + throughout we will be working in a relational language @xmath16 with only one ternary relation @xmath17 finite @xmath0-structures are denoted by @xmath18 and by @xmath19 we mean some arbitrary @xmath0-structures .
by @xmath20 we mean that @xmath21 is a finite substructure of @xmath22 by @xmath23 we denote a class of finite @xmath0-structures , always presumed to contain @xmath24 for a structure @xmath25 by @xmath26 we denote the class of its finite substructures . for a given @xmath27 by @xmath28
we denote the class of all @xmath0-structures @xmath29 with @xmath30 finally for @xmath31 the structure induced by @xmath32 on @xmath33 is denoted by @xmath34 we begin by recalling some of the basic definitions in the context of _ smooth classes _ , and for a thorough investigation of smooth classes we refer the reader to @xcite , , , and @xcite . unless otherwise stated , by @xmath35 we mean an arbitrary smooth class and all the structures are assumed to be an element of @xmath36 [ dfnfreejoin ] for every @xmath37 with @xmath38 + ( i ) we say that @xmath39 is _ free _ over @xmath21 with respect to ( or from ) @xmath40 if there is no relation with at least one node in @xmath41 and at least one node in @xmath42 + ( ii ) the structure @xmath43 is called the _ free join _ of @xmath39 and @xmath32 over @xmath21 , denoted by @xmath44 if the universe of @xmath43 is @xmath45 and we have : @xmath46 [ dfnintext ] suppose that @xmath47 + ( i ) we say that @xmath39 is a @xmath48-_intrinsic extension _ of @xmath49 if for any @xmath32 with @xmath50 we have that @xmath51 in this situation we write @xmath52 + ( ii ) we say that @xmath53 is a _
minimal pair _ , in notation @xmath54 if for any @xmath32 with @xmath50 we have @xmath55 but @xmath56 [ dfnoldcl ] suppose that @xmath57 .
we say that @xmath32 is @xmath48-_old _ over @xmath21 with respect to @xmath39 whenever @xmath58 and @xmath32 is free over @xmath21 with respect to @xmath59 otherwise we say that @xmath32 is @xmath48-_new _ over @xmath21 with respect to @xmath59 it follows from the definition above that whenever @xmath57 and @xmath32 is free over @xmath21 with respect to @xmath60 then @xmath58 .
the following lemma is standard and follows easily from the definitions above : [ lmaintclunion ] let @xmath61 the following hold in @xmath29 : + ( i ) if @xmath62 and @xmath63 , then @xmath64 + ( ii ) if @xmath65 and @xmath66 , then @xmath67 + ( iii ) if @xmath65 and @xmath66 , then @xmath68 + ( iv ) if @xmath69 then @xmath52 + ( v ) if @xmath70 then there is a finite sequence @xmath71 such that @xmath72 and @xmath73 [ dfngenericmodel ] a countable @xmath0-structure @xmath29 is called @xmath35__-generic _ _ if the following conditions hold : + ( 1 ) @xmath74 + ( 2 ) whenever @xmath75 and @xmath76 , there is @xmath77 such that @xmath78 and @xmath79 + ( 3 ) @xmath29 is the union of a chain @xmath80 where @xmath81 and @xmath82 for each @xmath83 [ dfnchi ] let @xmath74 @xmath84 and @xmath85 then : + ( i ) by a _ copy _ of @xmath39 over @xmath21 in @xmath29 , we mean the image of an embedding of @xmath39 over @xmath21 into @xmath22 + ( ii ) by @xmath86 we mean the number of distinct copies of @xmath39 over @xmath21 in @xmath29 ( which can be infinite ) .
[ dfnclosure ] for any @xmath87 and @xmath20 we define @xmath88 [ factclosureproperties ] ( i ) @xmath89 is the unique minimal structure containing @xmath21 closed in @xmath22 + ( ii ) if @xmath29 is @xmath35-generic , then the closure of each finite substructure of @xmath29 is finite .
[ dfnac ] a smooth class @xmath23 is said to have the _ algebraic closure property _ , in short _ ac _ , if there is a function @xmath90 such that for any @xmath91 with @xmath92 and any @xmath87 we have @xmath93 if a class @xmath23 has ac , then for any @xmath94 and @xmath20 , @xmath95 is finite . the notion of genericity for a smooth class @xmath35 is not necessarily first order expressible , but there exists a first order analogue , which is called _ semigenericity _ first introduced in for smooth classes with ac .
it was then generalized to a broader context by the second author in @xcite and @xcite .
[ dfnsemigenericmodel ] an @xmath0-structure @xmath29 is called @xmath35__-semigeneric _ _ if : + ( 1 ) @xmath74 + ( 2 ) for any @xmath20 , any @xmath39 with @xmath76 , embedding @xmath96 , and @xmath97 , there is an embedding @xmath98 that extends @xmath99 and @xmath100 that is , .
the reader is referred to , to see that the notion of semigenericity can be expressed by a first order theory .
a vast variety of smooth classes is obtained from the class of finite structures equipped with certain functions called _ _ predimension__s .
[ dfnpredimension ] a function @xmath101 is called a _ predimension _
if for any @xmath91 we have + ( 1 ) @xmath102 + ( 2 ) if @xmath103 then @xmath104 + ( 3 ) @xmath105 a given class @xmath23 can be equipped with several predimension functions , but we will be interested in ones which are proved to have interesting model theoretic properties . for a real number @xmath106 $ ] define @xmath107 as @xmath108 where by @xmath109 we mean the number of relations in @xmath21 .
we assume that @xmath1 is symmetric and realized only with distinct triples of elements .
based on these predimensions one can define two _ strong _ submodel relations on @xmath23 that are denoted by @xmath110 and @xmath111 [ dfnsmoothclassexamples ] for @xmath91 we write + ( i ) @xmath112 if for any @xmath113 with @xmath114 we have @xmath115 + ( ii ) @xmath116 if for any @xmath113 with @xmath117 we have @xmath118 + ( iii ) also let @xmath119 and @xmath120 + if @xmath112 ( or @xmath121 ) we say that @xmath21 is @xmath110-strong ( or @xmath122-strong ) in b. the corresponding intrinsic extensions for the classes @xmath123 and @xmath124 are denoted respectively by @xmath125 and @xmath126 in these classes the notion of intrinsic extension can be described concretely as follows .
[ factintrisicforpredim ] ( i ) @xmath127 if and only if for any @xmath32 with @xmath50 we have @xmath128 + ( ii ) @xmath129 if and only if for any @xmath32 with @xmath50 we have @xmath130 both @xmath131 and @xmath132 are smooth classes with the full amalgamation property possessing the properties necessary for the existence of a generic structure .
hence we reserve @xmath133 and @xmath6 respectively for the generic structures of these two classes .
also set @xmath134 and @xmath135 denote by @xmath136 and @xmath137 respectively the theory of the semigeneric structures of @xmath131 and @xmath138 notice that since full amalgamation holds for these two classes , their corresponding generics satisfy respectively @xmath136 and @xmath139 also we let @xmath140 and @xmath141 denote the corresponding closures .
it is worth noting that if @xmath142 is an irrational , then the classes @xmath123 and @xmath124 are the same and for any finite @xmath143 the relation @xmath144 is equivalent to @xmath145 moreover by lemma 3.19 and 3.22 of , we know that for every @xmath5 , $ ] the class @xmath131 has ac ; the reason is that whenever @xmath146 the relative predimension of @xmath39 over @xmath21 is negative and each additional copy of @xmath39 strictly decrease the predimension and hence for any @xmath74 the number @xmath86 is bounded above .
in addition , we refer to the proof of lemma 3.19 and 3.22 in , for the reason why this upper bound is uniformly determined by @xmath147 and @xmath148 on the other hand the class @xmath149 for rational @xmath142 , does not have ac , the reason is that for a structure @xmath150 there are intrinsic extensions @xmath39 with @xmath151 this fact allows the copies of @xmath39 over @xmath21 to be infinite .
but even in this class , @xmath86 is uniformly bounded for intrinsic extensions with @xmath152 in such cases being @xmath153-intrinsic extension implies @xmath125-intrinsic extension and we have the following fact , which is the same as lemma 3.19 in but it is rephrased in a way which is more suitable for the present context : [ factupperbound ] for every @xmath5 , $ ] there is a function @xmath154 such that for every @xmath155 and every @xmath156 we have @xmath157 moreover @xmath158 where by @xmath159 we mean the set of relations with at least one node in @xmath41 and at least one node in @xmath160 in order to draw a dividing line in terms of the classes @xmath131 and @xmath132 that are the typical examples of smooth classes with and without ac , we collect some results about these classes .
an important feature here is that the types of tuples in the models of the theory of generic structures are expected to be determined by their closures .
hence one can see that in the presence of ac , that bounds the size of the closures , the smooth class leads to a stable context .
[ factstablecontext ] ( i ) for every @xmath5 , $ ] we have that @xmath161 moreover @xmath162 is a near - model - complete but not model - complete theory ( ) .
+ ( ii ) if @xmath163 then @xmath162 is @xmath164-stable ( ) .
+ ( iii ) if @xmath165 then @xmath162 is strictly stable ( ) .
furthermore @xmath162 is equal to @xmath166 the almost - sure theory of random graphs with edge probability @xmath167 hence it has the finite model property ( ) . but for @xmath149 for @xmath142 a rational , the situation for the theory of generic changes dramatically : [ factunstablecontext ] ( i ) @xmath137 interprets robinson arithmetic and is essentially undecidable ( ) .
+ ( ii ) all finite graphs are interpretable in @xmath6 and hence @xmath168 is undecidable and has the strict order property ( ) . also for
a smooth class @xmath23 one can associate a theory @xmath169 in an expanded language @xmath170 which reduces the notion of strong submodel in @xmath0 to the notion of @xmath171-substructure .
precise definitions and more investigations can be found in @xcite and @xcite .
it is also shown in @xcite and @xcite that the class of existentially closed models of @xmath172 is rather well - behaved .
in particular it is shown that this class is simple in the sense of @xcite .
recall that if @xmath173 is irrational , then the classes @xmath131 and @xmath132 are the same , and moreover the theory of their generic structure is near - model - complete . for rational @xmath142 , near - model - completeness still holds for @xmath131-generic @xmath174 in this section we prove that for a rational @xmath5 , $ ] the theory of the @xmath132-generic @xmath2 admits quantifier elimination down to a certain class of formulas , called _ closure formulas .
_ also as an application , we investigate the theory of a certain ultraproduct of generic structures .
for the rest of this section we fix a rational @xmath142 and to ease the notation , we drop the subscript @xmath142 from @xmath175 and @xmath176 and will respectively write @xmath177 and @xmath178 instead .
closure formulas ( [ dfnclosureformula ] ) are roughly speaking @xmath0-formulas @xmath179 in which all the quantifiers range through the intrinsic extensions of @xmath180 the closure formulas are inductively built through the following hierarchy .
let : @xmath181 and define @xmath182 and @xmath183 to be the sets which are obtained respectively from @xmath184 and @xmath185 by closure under conjunctions and disjunctions .
for @xmath186 denote by @xmath187 the formula that expresses the diagram of @xmath39 over @xmath21 .
we make the convention that @xmath188 are variables associated to the elements of @xmath21 and @xmath189 are variables containing @xmath41 and not intersecting @xmath180 we also allow @xmath190 to contain dummy variables . for any @xmath191 define @xmath192 to be the following set of formulas : @xmath193 and similarly @xmath194 to be the set : @xmath195 also define @xmath196 and @xmath197 respectively as closures under positive boolean operations of @xmath192 and @xmath198 finally consider the following sets : @xmath199 [ dfnclosureformula ] by a _ closure formula _ we mean a formula in @xmath200 a closure formula is called _ basic _ , if it lies in @xmath201 to avoid confusion , for every basic closure formula @xmath179 sometimes we have added @xmath53 as subscript to @xmath202 hence we display it by @xmath203 to emphasise on the initial diagrams appearing in @xmath204 for the proof of , we need to work with the diagrams that appear in basic closure formulas .
to each basic closure formula @xmath205 we associate a tree @xmath206 as in .
for such a formula @xmath203 its tree @xmath207 is comprised of a pair @xmath208 and is defined inductively as follows . [ dfntree ] for @xmath209 define @xmath210 for @xmath211 if the formula @xmath212 is of the form of the elements in ( [ eqnscn ] ) or ( [ eqnpcn ] ) define @xmath213 we call @xmath21 the root of @xmath207 and denote by @xmath214 the number of the leaves of @xmath215 one can easily check that there is a one - to - one correspondence between trees and formulas . the next remark and lemmas
demonstrate some properties of closure formulas and propose a more canonical way to describe them .
they also provide tools needed for proving .
[ rmrkbasicfactsaboutclosureformula ] ( i ) @xmath216 for any @xmath217 moreover @xmath218 + ( ii ) every @xmath219-formula is equivalent either to an antilogy or to a formula of the form @xmath220 for some @xmath221 + ( iii ) every @xmath222-formula is equivalent to a tautology , an antilogy , or to a formula of the form @xmath223 for some @xmath221 [ lmahierarchycanonicalformsigma ] ( i ) if @xmath224 and @xmath225 then @xmath226 is logically equivalent to a formula of the following form : @xmath227 where @xmath228 and for each @xmath229 @xmath230 where @xmath231 hence a conjunction of @xmath232-formulas is equivalent to a disjunction of @xmath233-formulas . + ( ii ) if @xmath234 and @xmath235 then @xmath226 is equivalent to a disjunction of @xmath192-formulas , where @xmath231 + ( iii ) let @xmath236 be a quantifier free @xmath0-formula implying @xmath237 for some @xmath221 then for any @xmath238 the following formula , @xmath239 , is a @xmath240-formula : @xmath241 if @xmath242 based on we can assume that @xmath243 so suppose that @xmath244 and @xmath245 are respectively as followings : @xmath246 if @xmath247 , it is easy to check that @xmath226 is logically equivalent to @xmath248 so suppose that @xmath249 and name this common structure @xmath160 then define @xmath250 and @xmath251 respectively to be @xmath252 and @xmath253 note that for a tuple @xmath254 there are actually four possible cases ; the first one is when neither @xmath255 nor @xmath256 are satisfied over @xmath257 the second case is when @xmath255 is satisfied over @xmath254 but @xmath256 is not .
the third case is when @xmath256 is satisfied over @xmath254 but @xmath255 is not . and
the last case , when @xmath255 and @xmath256 are both satisfied over @xmath257 it can be easily seen that in the first three cases , @xmath226 is respectively equivalent to @xmath258 or @xmath259 now in the last case , note that by we have @xmath260 notice that there is a finite number of possible diagrams @xmath261 for @xmath262 to be realized over @xmath263 and therefore the conjunction @xmath226 is logically equivalent to : @xmath264,\end{aligned}\ ] ] which introduces the required @xmath265s .
\(ii ) by essentially a similar argument to ( i ) .
\(iii ) first note that the formula @xmath239 is equivalent to : @xmath266 and for a set @xmath32 with @xmath267 there exist finitely many complete diagrams which have @xmath45 as their universe and extend @xmath39 as their substructure .
enumerate them by @xmath268 and notice that since the language is finite relational , is equivalent to : @xmath269 therefore by part ( i ) of , this formula is a @xmath240-formula .
[ crlcanonicalform ] every closure formula @xmath270 is equivalent to a disjunction of @xmath271-formulas of the following form @xmath272 where @xmath273 and for each @xmath274 we have @xmath275 notice that in , in fact a combinatorial measurement " was derived for every @xmath5 $ ] in such a way that it decides the satisfaction of a given formula @xmath179 uniformly in all models of @xmath276 more precisely , since ac holds in @xmath131 , there exists a natural number @xmath277 such that for any @xmath278 and any @xmath279 the satisfaction of @xmath280 depends only on the isomorphism type of @xmath281 again @xmath282 thanks to ac , is finite ( uniformly in all models of @xmath136 ) and hence @xmath136 is both complete and near - model - complete . on another basis ,
in our case we propose a measurement in terms of logical complexity " which relates the satisfaction of @xmath179 to a formula which describes a bounded part of the closure of @xmath180 to introduce this measurement a machinery needs to have appeared before , that works even in the absence of ac .
more precisely , for any @xmath179 one can find a natural number @xmath283 and a closure formula in @xmath284 that is equivalent to @xmath179 under @xmath285 the next definition , distinguishes the non - ac from the ac part inside @xmath286 in this definition we highlight a distinction among closure formulas , and introduce a dividing line one side of which contains the formulas with a finite nature as in the classes with ac .
in fact this _ non - primary _ closure formulas are the ones constructed from @xmath287-intrinsic extensions .
but the other side of the dividing line contains the _ primary _ formulas that in the classes without ac , constitute the noteworthy part of closure formulas .
it can be subsequentially seen in the proof of how the non - primary formulas collapse to the lower levels of the hierarchy , namely to @xmath288-formulas .
the definition of primary formulas is done by induction on the levels of @xmath200 [ dfnprimaryclosureformula ] we define _ primary _ formulas as follows : + ( i ) both @xmath185 and @xmath222-formulas are _
primary_. for @xmath289 the formula @xmath290 ( which is of the form of the elements of ( [ eqnpcn ] ) ) is primary if @xmath291 and for each @xmath292 we have @xmath293 + ( ii ) every @xmath184-formula is primary . for @xmath211 the formula @xmath294 ( which is of the form of the elements of ( [ eqnscn ] ) )
is primary if for each @xmath292 the formula @xmath295 is a primary @xmath296-formula .
+ towards simplifying formulas dealt with in the proof of , next lemma enables us to further reduce @xmath271-formulas to the primary ones .
the proof of this lemma suggests a useful interaction between the notions of @xmath297 and @xmath298 in fact the idea is to eliminate the negative part of @xmath299-intrinsic extensions , namely the @xmath297-part , by replacing them with existential closure formulas .
the adequacy of this approach is guaranteed by which introduces a uniform bound on the number of copies of @xmath287-intrinsic extensions .
[ lmaprimaryformulas ] every closure formula is equivalent to a disjunction of primary @xmath271-formulas . according to and we only need to show that any @xmath194-formula @xmath203 is equivalent to a disjunction of primary @xmath271-formulas .
a formula @xmath290 is of the form of the elements of ( [ eqnpcn ] ) .
we first show that we may assume that @xmath300 so suppose on the contrary that @xmath301 then define @xmath302 and let @xmath303 therefore @xmath304 and @xmath305 also it is obvious that @xmath306 now by there is a function @xmath307 which uniformly bounds the number of copies of @xmath308 over @xmath160 set @xmath309 the formula @xmath310 is equivalent to the following formula @xmath311,\ ] ] with @xmath312 and @xmath313 moreover @xmath314 can be obtained by a finite disjunction of existential closure formulas .
now by an easy induction on @xmath315 without loss of generality we can assume that @xmath300 now consider that in @xmath310 ( which is in the form of the elements of ( [ eqnpcn ] ) ) each @xmath295 is of the following form @xmath316 where @xmath317 enumerate @xmath318s by @xmath319 in such a way that for each @xmath320 we have @xmath321 and for each @xmath322 we have @xmath293 now define for each @xmath320 the following structures @xmath323 and note that @xmath324 and @xmath325 hence @xmath326 since @xmath327 we have @xmath328 moreover notice that @xmath329 since otherwise if @xmath330 is the number of the relations that avoid this free amalgamation , then we have @xmath331 which contradicts the fact that @xmath332 the fact that @xmath333 is free from @xmath334 over @xmath335 implies that for any @xmath320 and any two realizations of @xmath39 over @xmath49 say @xmath336 and @xmath337 with @xmath338 a realization of @xmath333 over @xmath336 guarantees the existence of a realization of @xmath333 over @xmath339 also since @xmath340 again by there is a function @xmath307 which uniformly bounds the number of copies of @xmath333 over @xmath160 let @xmath341 now it is easy to check that @xmath310 is equivalent to the following formula @xmath342,\ ] ] where @xmath343 is as follows @xmath344\big).\ ] ] note that @xmath345 can be expressed by a finite disjunction of complete diagrams .
the crucial part of the induction that is used in the proof of is to demonstrate for a closure formula @xmath346 that formula @xmath347 also lies in @xmath348 the idea of the proof is to extract from the diagrams appearing in @xmath349 the maximal possible part that can be expressed by a closure formula , say @xmath350 and to show that the satisfaction of @xmath351 implies the satisfaction of @xmath352 in fact for a given @xmath353 satisfying @xmath350 the genericity of @xmath354 enables us to find a suitable @xmath355 that satisfies @xmath356 the following technical lemma is required for defining @xmath357 [ lmanegativecore ] let @xmath358 with @xmath359 and @xmath360 then there exists a finite number of structures @xmath361 and a finite number of @xmath182-formulas @xmath362 with @xmath363 and @xmath364 for each @xmath365 such that the following hold : @xmath366 for any @xmath367 we have that @xmath368 and the formula @xmath369 implies @xmath370 in @xmath371 @xmath366 for any embedding @xmath372 there exist some @xmath373 such that @xmath374 ( witnessed by some @xmath375 ) and for every embedding @xmath376 with @xmath377 we have @xmath378 first note that there are finite number of possibilities @xmath379 such that @xmath380 and having @xmath381 not free over @xmath382 with respect to @xmath32 for each @xmath383 hence for each @xmath384 we have @xmath385 and @xmath386 let @xmath387 and notice that @xmath388 and @xmath389 thus by we have that @xmath390 for some natural number @xmath391 therefore one can find a natural number @xmath392 and pairs @xmath393 satisfying the following conditions : @xmath366 for any @xmath394 we have @xmath395 and @xmath396 is obtained by a union of copies of @xmath43 over @xmath382 with @xmath397 also @xmath398 is a family of nonempty substructures of @xmath399 @xmath366 for any embedding @xmath372 there exist some @xmath400 and an embedding @xmath401 with @xmath402 such that for any copy @xmath403 of @xmath43 over @xmath404 we have that @xmath405 if and only if there exists a copy @xmath406 of @xmath43 over @xmath407 such that @xmath408 now one can see that for any embedding @xmath372 one of the structures @xmath396 is realized over @xmath409 and for every embedding @xmath376 its image @xmath410 is realized free from @xmath411 over @xmath404 if and only if @xmath412 clearly the latter property is expressible by a @xmath182-formula which introduces the desired @xmath413 to finish the proof , note that for any @xmath400 there are finitely many complete diagrams over @xmath414 extending both @xmath32 and @xmath399 therefore by assigning the same @xmath369 to all the possible diagrams of @xmath415 we may get a new list @xmath416 for some @xmath417 as required . [ [ definition - of - phi_psibarx . ] ] definition of @xmath357 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for any @xmath418 we define a closure formula @xmath351 ( that will be proved to be equivalent to @xmath347 in )
. the definition is given by induction on @xmath419 suppose that @xmath420 hence for some @xmath421 it is of the form of @xmath422 set @xmath423 and let @xmath351 be @xmath424 for @xmath191 suppose that the innermost @xmath425-formulas appearing in @xmath426 are of the following form @xmath427 in which @xmath428 moreover by induction we may assume that @xmath429 and @xmath430 now by cutting all the leaves and their immediate predecessors from @xmath431 we obtain a tree @xmath432 which corresponds to a formula @xmath433 this formula belongs to a lower level of the hierarchy and in fact we have @xmath434 hence by induction hypothesis one can define a closure formula @xmath435 we may further assume by induction that the innermost @xmath219-formulas appearing in @xmath436 are formulas of the following form @xmath437 where @xmath438 let us denote these formulas by @xmath439 now in order to define @xmath351 we replace each formula @xmath440 in @xmath436 with a new formula @xmath441 defined as follows .
let @xmath442 and for @xmath443 select those @xmath444s which are and renumerate them by @xmath445 we have two different cases : * case 1 , * in this case let @xmath441 be : @xmath446 * case 2 , * in this case for each @xmath447 set @xmath448 now obtain @xmath449 by applying to @xmath450 and @xmath451 and for any @xmath373
set @xmath452 moreover let @xmath453 be the following formula @xmath454\big).\end{aligned}\ ] ] since @xmath455 implies @xmath456 by part ( iii ) of the above formula is a @xmath457-formula .
now let @xmath441 be the following formula @xmath458 [ thmqe ] @xmath459 admits quantifier elimination down to closure formulas .
more precisely , each @xmath460-formula @xmath461 is equivalent to a closure formula in @xmath354 .
we proceed by induction on the complexity of @xmath461 . the induction base for atomics as well as the induction steps for negation , conjunction and disjunction are obvious .
therefore based on we may suppose that @xmath461 is of the form of @xmath462 where @xmath463 is a primary @xmath192-formula of the form of .
now by induction on @xmath315 we show that @xmath464 to see that @xmath465 implies @xmath350 using the notations employed in defining @xmath350 note that for any @xmath466 we have that @xmath467 also by induction hypothesis we know that @xmath468 implies @xmath469 but since each innermost @xmath457-formula appearing in @xmath426 logically implies the corresponding introduced @xmath441 , we have that @xmath470 for the other direction , assume for some @xmath353 that @xmath470 to ease the notation we omit subscript @xmath354 from @xmath471 and write @xmath472 instead .
[ [ claim . ] ] claim .
there is a structure @xmath473 satisfying the following properties : @xmath474 @xmath366 for any strong embedding of @xmath475 into @xmath354 over @xmath476 there exist some @xmath477 with @xmath478 we define the structure @xmath475 by induction on @xmath315 . for @xmath479 let @xmath475 be the structure @xmath480 and for @xmath481 define it to be @xmath482 for @xmath191 following the notations used in the definition of @xmath350 we obtain the formula @xmath483 and a tree @xmath434 now let @xmath484 be an enumeration of the diagrams appearing as the leaves of @xmath485 by induction hypothesis we know that @xmath486 is already defined .
note that corresponding to each @xmath318 there is a @xmath457-formula @xmath487 in the form of . to avoid any ambiguity we attach an @xmath488 at the beginning of all the subscripts used in the definition of @xmath357 now for each @xmath489 enumerate by @xmath490 the actual realizations of @xmath491 over @xmath333 in @xmath371 moreover
suppose for each @xmath492 that @xmath493 is one of the structures @xmath494 which is forced by @xmath495 to be realized over @xmath496 let @xmath497 be the structure @xmath498 forced by @xmath495 to be realized over @xmath254 as a subset of @xmath476 and consider @xmath499 to be a structure isomorphic to @xmath500 then for each @xmath492 let @xmath501 be a structure which is isomorphic to @xmath502 and define @xmath503 to be the following structure @xmath504 in which @xmath501s are mutually freely joined over @xmath499 and for each @xmath505 we have @xmath506 and @xmath507 denote by @xmath508 the structure whose universe is @xmath509 in which all @xmath503s are mutually freely joined over @xmath472 having that @xmath510 now we show that @xmath511 for each @xmath384 and @xmath505 suppose that @xmath512 and @xmath513 be arbitrary subsets of @xmath499 and @xmath501 respectively , and note that according to the way @xmath508 is defined we have the following equations @xmath514 but for each @xmath384 we have @xmath515 moreover @xmath516 and because @xmath517 and @xmath518 we have @xmath519 now notice that @xmath520 and since @xmath426 is considered to be primary , for each @xmath384 and @xmath521 we have @xmath522 hence for each @xmath505 we have @xmath523 therefore we have @xmath524 and hence @xmath525 now set @xmath526 finally since @xmath354 is generic , @xmath475 can be strongly embedded over @xmath527 the existence of a tuple @xmath528 with the property that @xmath529 is obvious by noticing that in the definition of @xmath350 for each @xmath384 we only take into account those @xmath491s that are and since we embed @xmath475 over @xmath472 in an old way , the @xmath491s that are not can not be satisfied at all .
the situation is similar for those @xmath530 for which we have @xmath531 an important difference between @xmath354 and an arbitrary semigeneric structure @xmath532 is that while the closure of any finite tuple is finite in the former , it can be infinite in the latter .
as one can see in the proof of , we strongly made use of this property inside @xmath371 hence a further clarification of the distinctions between @xmath533 and @xmath534 can be obtained by answering the following question .
does @xmath535 admit a quantifier elimination down to closure formulas ?
the next corollary shows that the type of a finite tuple is completely determined by its closure .
this corollary also generalizes corollary 2.4 of @xcite and lemma 1.30 of to the context of @xmath286 [ crltype ] suppose that @xmath29 is a saturated model of @xmath285 let @xmath536 both have a same diagram @xmath160 then we have : @xmath537 we prove that having isomorphic closures implies equality of types , and the other direction is rather obvious .
so note that if @xmath538 then for any @xmath97 and any @xmath539 we have that @xmath540 hence by we have @xmath541 in the rest of this section we show that is true for a certain ultraproduct of models each of which is elementarily equivalent to a generic structure . using this result
we prove that the theory of this ultraproduct is the same as @xmath542 in fact it is proved in that if @xmath543 $ ] is an strictly decreasing sequence of real numbers converging to a rational @xmath5 , $ ] and for each @xmath97 we consider a structure @xmath544 elementarily equivalent either to @xmath545 or to @xmath546 and @xmath547 is a non - principal ultrafilter on @xmath548 then the ultraproduct @xmath549 is a model of @xmath139 we show that holds in @xmath550 using which we prove that @xmath549 is elementarily equivalent to @xmath551 in , for each @xmath142 we introduced an upper bound on @xmath552 say @xmath553 accordingly , it is easily seen that for every @xmath554 $ ] we have @xmath555 hence in particular for a strictly decreasing sequence @xmath556 converging to some @xmath5\cap\mathbb{q } , $ ] if @xmath557 and @xmath558 then there exists a natural number @xmath559 such that for every @xmath560 we have @xmath561 and @xmath562 moreover for any @xmath560 we can observe that @xmath563 this phenomena seems natural considering the fact that @xmath564 is actually infinity . note that the notions of intrinsic extension and strong embedding change from @xmath124 to @xmath565s
. therefore we denote by @xmath566 the set of closure formulas defined for @xmath565 and @xmath567 however we have the following lemma .
[ lmaseqprop ] ( i ) for any structure @xmath568 there exists a natural number @xmath569 such that for any @xmath570 we have @xmath571 + ( ii ) for any structures @xmath572 there exists a natural number @xmath573 such that for any @xmath574 we have @xmath575 and hence @xmath576 + ( iii ) for any structures @xmath577 there exists a natural number @xmath578 such that for any @xmath579 we have @xmath580 and hence @xmath581 [ crlseqclosure ] for any structures @xmath582 there is a natural number @xmath583 such that for any @xmath584 we have @xmath585 [ thmproductqe ] ( i ) suppose that there exists a natural number @xmath583 such that for any @xmath584 we have @xmath586 and @xmath544 is elementary equivalent to the @xmath587-generic .
then for each formula @xmath588 there exists a closure formula @xmath589 such that @xmath590 + ( ii ) @xmath591 \(i ) by for each @xmath592 there is a closure formula @xmath589 such that @xmath593 recall that the main step in proving amounts to defining the closure formula @xmath351 and proving the expression ( [ eqnmainpartoftheorem ] ) .
now consider a given formula @xmath179 in the form of @xmath347 with @xmath594 first note that based on there is a natural number @xmath595 such that for any @xmath596 we have @xmath597 moreover based on there exists a natural number @xmath598 such that the closure of structures in @xmath565 remains unchanged for all @xmath599 hence for all @xmath600 the procedure of defining @xmath351 leads to the same closure formula as the closure formula we obtain for @xmath148 furthermore holds for an arbitrary fixed @xmath5\cap\mathbb{q}. $ ] so there exists a natural number @xmath601 such that for any @xmath602 we have @xmath603 therefore @xmath604
( ii ) let @xmath29 and @xmath605 be respectively saturated models of @xmath168 and @xmath606 based on part ( i ) and using compactness it is easy to see that the following set defines a back and forth system between the substructures of @xmath29 and @xmath607 @xmath608 hence we have @xmath591 note that the hypothesis in the first part of the theorem can be replaced by the following statement : `` for almost all @xmath97 we have that @xmath586 and @xmath544 is elementary equivalent to the @xmath587-generic '' .
in this section we show that @xmath168 does not have the finite model property , for each @xmath5\cap\mathbb{q}.$ ] towards this end we give a weaker form of proposition 3.3 of which suffices for our purpose . throughout this section
we omit the subscript @xmath148 in fact this lemma is restricted to @xmath609 instead of @xmath535 : [ lmadefinfinrel ] for any @xmath610 there is a definable relation @xmath611 , symmetric in the first @xmath521 variables , such that for any @xmath612 and any @xmath613^{k } $ ] there exist some @xmath614 such that for any @xmath353 we have @xmath615 [ rmrkbasicoperation ] it is easy to see that one can find new codes for the sets obtained by the basic set - theoretic operations on some given coded sets . to ease later references
we fix a notation for the formulas which correspond to some of these operations .
so suppose that @xmath620 and @xmath621 code respectively the sets @xmath622^{k } $ ] and @xmath623^{k } $ ] with @xmath624 now we fix the following : fix three finite substructures @xmath638 and @xmath32 of @xmath354 with @xmath639 , @xmath640 , @xmath641 , and @xmath642 the idea is to interpret a positive rational number @xmath643 by a copy of @xmath49 over which it is realized @xmath644 copies of @xmath60 and @xmath645 copies of @xmath646 in fact @xmath647 plays the role of the numerator and @xmath648 that of the denominator .
we borrow the notion of a basis from .
so for each @xmath655 let @xmath656 and @xmath657 respectively denote the union of all copies of @xmath39 and @xmath32 over @xmath257 a subset @xmath658 is called a @xmath39__-basis _ _ for @xmath254 if it contains exactly one element of each copy of @xmath39 over @xmath254 .
similarly we can define a @xmath32__-basis__. the notion of @xmath39-basis can be expressed by a formula @xmath659 defined as following @xmath660.\end{aligned}\ ] ] in fact @xmath659 indicates that @xmath661 codes a @xmath39-basis for @xmath180 similarly we may set a formula @xmath662 to express the notion of @xmath32-basis .
the equivalence relation we consider on @xmath354 is defined in such a way that each @xmath663 and @xmath664 in @xmath354 be equivalent if and only if @xmath665 so let @xmath666 be the following formula @xmath667,\end{aligned}\ ] ] where @xmath668 the formula above expresses literally that while @xmath669 and @xmath670 code respectively a basis for @xmath671 and @xmath672 , say @xmath673 and @xmath674 , @xmath675 and @xmath621 define respectively a basis for @xmath676 and @xmath677 , say @xmath678 and @xmath679 .
furthermore @xmath680 and @xmath681 code respectively @xmath682 and @xmath683 and @xmath684 defines a bijection between @xmath682 and @xmath685 . note that using the genericity of @xmath686 for any @xmath687 with @xmath688 and @xmath689 the disjoint union of the representatives of the classes @xmath643 and @xmath690 is strongly embeddable into @xmath691 hence the formulas introduced in have still their valid meanings .
also since the closures of finite subsets of @xmath354 are finite , it is easy to check that the above formula defines an equivalence relation with the required property . to define an order on @xmath692 it suffices to replace the bijection used in @xmath666 by an injection .
therefore let @xmath693 be the following formula @xmath694.\end{aligned}\ ] ] where @xmath668 now it is easy to check that @xmath695 interprets a linear order .
moreover since @xmath696 , for any @xmath697 the predimension of the structure constructed by @xmath21 with exactly @xmath644 disjoint copies of @xmath39 , all be mutually freely joined over @xmath21 , is strictly positive and hence this structure lies in @xmath698 .
the same is true for any @xmath699 with @xmath688 and the structure constructed by @xmath21 with exactly @xmath644 disjoint copies of @xmath39 and @xmath645 disjoint copies of @xmath32 all be mutually freely joined over @xmath21 .
also note that all of these structures are embedded strongly in @xmath371 therefore @xmath695 is dense and we have @xmath700 acknowledgement .
the first author greatly benefited from discussions on this paper with john baldwin and his comments .
we would like to thank zaniar ghadernezhad for some helpful discussions .
also we would like to thank mohsen khani for carefully reading the paper and giving useful comments and suggestions .
|
in the context of hrushovski constructions we take a language @xmath0 with a ternary relation @xmath1 and consider the theory of the generic models @xmath2 of the class of finite @xmath3-structures equipped with predimension functions @xmath4 for @xmath5\cap\mathbb{q } $ ] .
the theory of generic structures of non - ac smooth classes have been investigated from different points of view , including decidability and their power in interpreting known structures and theories . for a rational @xmath5 , $ ]
first we prove that the theory of @xmath6 admits a quantifier elimination down to a meaningful class of formulas , called _ closure formulas _ ; and on the other hand we prove that @xmath7 does not have the finite model property .
|
the explore ( extrasolar planet occultation research ) project is a series of searches for transiting extrasolar planets orbiting galactic plane stars using 4-m class telescopes . as an integral part of the search strategy , radial velocity ( rv )
follow - up observations for mass confirmation are done on 810 m class telescopes . in june 2001
, we used the ctio 4-m telescope for 11 nights ( 6 clear ) to observe the explore i field , located at @xmath0 ( malln - ornelas et al.2002 ) .
the best 37,000 light curves were examined , and rv follow - up of three planet candidates was done on the vlt in september 2001 ( malln - ornelas et al . , in prep . ) . in december 2001
we used the 3.6-m cfht for 16 nights ( 14 clear ) to observe the explore ii field , located at @xmath1 ( yee et al .
, in prep . ) .
the best 12,000 light curves were examined , and rv follow - up of 2 planet candidates and 2 additional eclipsing systems was done at keck in february 2002 ( malln - ornelas et al . , in prep . ) . here
we present a summary of the transiting planet candidates for which we have rv follow - up .
light curves with very precise photometry and good time sampling can be used to determine the eclipse shape and select the best transiting planet candidates with minimal contamination from other systems that can mimic the transit signature . in general terms , the best planet candidates will have very shallow eclipses ( @xmath223% , implying a small companion ) of short duration ( @xmath23h for a 34 day period transit , implying a dwarf star ) , with a clear flat bottom , and preferrably with a steep ingress / egress ( i.e. , a `` box shape '' ) .
for a full discussion on candidate selection and possible contaminants see seager & malln - ornelas ( 2002 ) and malln - ornelas et al.(2002 ) . a mass measurement from rv observations is required in order to confirm the presence of a planet , as opposed to a brown dwarf or a stellar companion .
in particular , the rv data should show only one velocity peak ( indicative of the presence of a single star ) , and ideally result in a mass detection ( with @xmath3 ) .
note that in the absence of an actual mass detection , a good upper limit of 12 @xmath4 may be used to make the case for a planet _ as long as the eclipses can be confirmed with certainty , and possible contaminants can be ruled out with confidence_.
a total of three planet candidates from the explore i search were followed - up with the vlt+uves in september 2001 . for the explore ii search , two planet candidates as well as two ( brigher ) stars with
deeper eclipses were observed with keck+hires in february 2002 . based on the window function of our observations , on
the number of light curves examined , assuming that 0.75% of single stars have a close - in giant planet , and adopting a binary fraction of 1/2 , we expect to find 1 planet in the explore i search and 12 planets in the explore ii search . here
we summarize our preliminary findings on the five planet candidates for which we have conducted rv follow - up observations .
the essential information is presented in tables 1 and 2 , and a brief discussion is given for each candidate .
a full description of our results will be presented in two future papers ( malln - ornelas et al .
, in prep . ) .
light curves and preliminary rv plots for some of the systems discussed here can be found in yee et al .
( 2002 ) .
table 1 lists the i and v magnitudes , eclipse depth , and period for the five transiting planet candidates with rv follow - up . note that in some cases it is possible that the period listed is a multiple of the true period , since additional intervening eclipses may have occurred during the day .
table 2 gives the eclipse and rv characteristics of the transit candidates , which can be used to determine whether each system is a good planet candidate .
_ as mentioned in 2 ,
the best planet candidates have `` box - shaped '' short - duration eclipses with clear flat bottoms . in the case of a planet
the rv should show only one velocity peak in the cross - correlation , and any rv variations should be of very small amplitude .
_ for some of the explore data , poor s / n in the light curve prevented us from ascertaining the eclipse shape in some cases , marked with a `` ? '' in table 2 .
bad weather during the rv follow - up resulted in poor phase coverage for ex2c11s4809 , which unfortunately prevented us from measuring a mass for this otherwise promising planet candidate .
star id&i&v & # of eclipses&eclipse depth&period + & ( mag)&(mag)&&(mag)&(days ) + exp1c02s46830&16.1&17.7&1&0.015 & ? + exp1c01s52805&16.2&17.9&2&0.03&2.2 + exp1c07s18161&17.6&19.4&2&0.025&3.8 + exp2c11s4809&18.3&20.0&3&0.017&3.0 + exp2c10s5069&18.4&20.1&4&0.008&4.0 + + star id&&&possible + & flat?&boxy?&short?&peaks&shifts&planet ?
+ exp1c02s46830&?&n&y&1&y & ? + exp1c01s52805&y&n&y&2&y&no +
exp1c07s18161&?&?&y&1&n&yes + exp2c11s4809&y&y&y&1&?&yes + exp2c10s5069&?&?&y&1&n&yes + @xmath5 * exp1c02s46830 . *
only one clear eclipse was detected , and the long ingress and egress duration make it an unlikely planet candidate . a steady decrease in rv of @xmath6500
m / s was detected over the 8 days spanned by the observations .
however , a planet can not be ruled out until a period is measured ( either from the eclipses or the rv variations ) which will enable a mass determination .
@xmath5 * ex1c01s52805 . *
the rv data for this object had two velocity peaks , revealing a system composed of a close eclipsing binary , plus a brighter superimposed star with no significant rv shifts .
note that the eclipses for this system have clear flat bottoms but _ are not _ `` box - shaped '' .
although it is possible for a planet crossing close to the stellar limb to produce transits with a long ingress and egress , such planet transits are much less common than `` box - shaped '' transits ( seager & malln - ornelas 2002 ) .
@xmath5 * exp1c07s18161 .
* this system has 2 detected eclipses , though the s / n is not high enough to ascertain eclipse shape with confidence .
there is only one rv peak , and we place a preliminary upper mass limit of 23 @xmath4 .
thus , this is a very promising planet candidate . to definitely establish this system as a planet , we require further photometric observations to confirm the eclipses , and ideally more rv data to obtain an actual mass detection .
@xmath5 * exp2c11s4809 .
* this system has very clear flat - bottomed eclipses , and only one rv peak .
however , due to poor weather we were unable to obtain enough phase coverage to place a useful mass limit .
further rv observations leading to a mass measurement are required to establish the nature of this system .
@xmath5 * exp2c10s5069 .
* we detected 4 eclipses , but all with very low s /
n so the eclipse shape could not be determined with certainty .
only one rv peak was found , with no detected shifts . to establish this system as a planet
we need more rv data to attempt a mass detection , as well as photometric observations to increase the s / n of the eclipses .
we have presented a status report on the transit candidates from the explore project s first two surveys . out of three systems with rv follow - up in the explore i search ,
one was found to be a close stellar binary plus a blended star , a second one is an unlikely planet candidate that nonetheless has not been entirely ruled out as a planet , and a third one is a promising planet candidate based on an upper mass limit .
the two candidates with rv follow - up in the explore ii search remain good planet candidates , although more rv data are required for both systems , and more photometric data are also desirable for exp2c10s5069 , which has extremely shallow eclipses detected with low s / n . we have shown that it is possible to produce a clean set of planet candidates from a deep transit search and place meaningful mass limits ( 23 @xmath4 ) on the companions via follow - up rv observations .
we emphasize that in order to find one good planet candidate , it is necessary to monitor several thousands of stars with good time coverage , high photometric precision , and time sampling which is good enough to determine the eclipse shape .
yee , h. k. c. , malln - ornelas , g. , seager , s. , gladders , m. d. , brown , t. , minniti , d. , ellison , s. l. , malln - fullerton , g. 2002 , to appear in proc .
spie conf . : astronomical telescopes and instrumentation , astro - ph/0208355
|
the explore project is a series of searches for transiting extrasolar planets using large - format mosaic ccd cameras on 4-m class telescopes .
radial velocity follow - up is done on transiting planet candidates with 810 m class telescopes .
we present a summary of transit candidates from the explore project for which we have radial velocity data .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
|
recently the increased interest in the properties of the @xmath3 and @xmath0 meson can be observed due to extensive experimental search of the @xmath3 and @xmath0 bound states performed e.g. at cosy @xcite , elsa @xcite , gsi @xcite , jinr @xcite , jparc @xcite , lpi @xcite , and mami @xcite as well as intensive theoretical investigations e.g. @xcite .
properties of @xmath0 in nuclear medium are related with the effects of @xmath4 anomaly at finite density @xcite , which is reflected in the large mass of the @xmath0 meson compared to the masses of the other members of the pseudoscalar meson nonet @xcite , and with the @xmath3-@xmath0 mixing @xcite .
experiment @xcite has provided already an important data for these studies @xcite , with the most precise direct measurement of the total width of the @xmath0 meson @xmath5@xcite , and the first rough estimation of the @xmath0-n interaction from the excitation function of the cross section for the @xmath6 reaction @xcite . here
we describe an analysis of the data used earlier for @xmath7 determination in view of the extraction of the production cross - section for the @xmath0 meson @xmath1 in proton - proton collisions and an update of the @xmath1 values presented previously @xcite .
in the reported measurement the @xmath0 meson was produced in proton - proton collisions reaction and its mass was reconstructed based on the momentum vectors of protons taking part in the @xmath8 reaction which was measured at five different beam momenta using the detector setup @xcite installed at the cooler synchrotron cosy @xcite in research centre jlich .
the schematic view of the detector setup is presented in fig .
[ fig : c11setup ] .
the collision of a proton from the beam with a proton cluster target may cause an @xmath0 meson creation . in that case
all outgoing nucleons have been registered by the detectors , whereas for the @xmath9 meson identification the missing mass technique was applied . the cosy beam momentum and
the dedicated zero degree facility enabled the measurement at an excess energy down the fraction of an mev above the kinematic threshold for the @xmath9 meson production .
modification of the target system allowed to decrease effective beam momentum spread and therefore enabled precise determination of the access energy @xmath2 with the precision of 0.10 mev .
good control of the systematic uncertainties was possible due to measurement performed at five different values of @xmath2 and monitoring of the beam and target properties @xcite . on the other hand the achieved missing mass resolution in the order of the total width of the @xmath0 meson itself @xcite improved significantly the @xmath0 production cross section measurement .
the number of registered @xmath0 mesons was obtained from the missing mass spectra for each @xmath2 value and corrected for the detector geometrical acceptance and registration efficiency .
the luminosity value was determined using comparison of the cross section of @xmath10 reaction determined by edda collaboration @xcite and the number of registered elastically scattered protons .
since @xmath1 measured at was obtained with the luminosity determination based on the edda data available that time @xcite , we updated these numbers accordingly to superseded data @xcite .
measurement at @xmath11 mev @xcite was already reported with new edda data @xcite , whereas spesiii @xcite and disto @xcite used different techniques for luminosity determination .
meson production in the proton - proton collision as a function of the access energy @xmath2 for the @xmath6 reaction measured at : ( solid circles - updated values of @xcite , solid square - measurement @xcite with usage of edda 2004 data @xcite ) , spesiii ( open squares ) @xcite and disto ( open triangle ) @xcite .
the solid line shows parametrization of the experimental data using formula [ eq : colin].,scaledwidth=72.0% ] the experimental data presented at fig .
[ fig : cross ] are compared to the analytical parametrization derived by fldt and wilkin @xcite which takes into account final state interaction of the protons : @xmath12 where @xmath2 denotes the excess energy , @xmath13 beam momentum , @xmath14 proton mass .
the parameters @xmath15 mev and @xmath16 mb denote the coulomb distortion and constant factor , respectively , and have been determined by fitting this formula to the experimental data .
values of @xmath6 cross sections determined at are gathered in the table [ tab : updated ] apart of the new measurement reported here , which is still in the final stage of the analysis .
.updated values of production cross - sections for the @xmath0 meson in proton - proton collisions measured at detector @xcite with statistical and systematic uncertainties , respectively .
[ cols=">,<,^ , > , < , > , < , < " , ]
this work has been supported by the polish national science center through grants no .
0320/b / h03/2011/40 , 2011/01/b / st2/00431 , 2011/03/b / st2/01847 , 2011/01/d / st2/00748 , 2011/03/n / st2/02652 , by the foundation for polish science through the project homing plus bis/2011 - 4/3 , by the european commission under the 7th framework programme through the ` research infrastructures ' action of the ` capacities ' programme ( fp7-infrastructures-2008 - 1 , grant agreement no . 227431 ) and by the ffe grants of the research center jelich .
00 p. moskal , j. smyrski , _ acta phys .
pol . _ * b41 * , 2281 ( 2010 ) .
m. skurzok , p. moskal , w. krzemien , _ prog .
* 67 * , 445 ( 2012 ) .
p. adlarson _ et al .
_ * c87 * , 035204 ( 2013 ) .
a. budzanowski _ et al .
, phys . rev . _ * c79 * , 012201 ( 2009 ) .
j. smyrski _ et al .
_ * b649 * , 258 ( 2007 ) . t. mersmann _ et al . ,
lett . _ * 98 * , 242301 ( 2007 ) .
m. nanova _ et al .
* b727 * , 417 ( 2013 ) .
y. k. tanaka _ et al .
, few body syst . _ * 54 * , 1263 ( 2013 ) .
k. itahashi _ et al .
phys . _ * 128 * , 601 ( 2012 ) .
afanasiev , _ phys .
_ * 8 * , 1073 ( 2011 ) .
h. fujioka , _ acta phys .
pol . _ * b41 * , 2261 ( 2010 ) .
h. fujioka , k. itahashi , _ hadron and nuclear physics _ * 09 * , 150 ( 2010 ) .
baskov _ et al .
_ : pos baldin - ishepp - xxi , 102 ( 2012 ) .
b. krusche _ et al .
, j. phys .
_ , * 349 * , 012003 ( 2012 ) . f. pheron _ et al . ,
* b709 * , 21 ( 2012 ) . c. wilkin , _ phys .
* b654 * , 92 ( 2007 ) . s.d .
bass , a.w .
thomas , _ phys .
lett . _ * b634 * , 368 ( 2006 ) . h. nagahiro _ et al .
rev . _ * c87 * , 045201 ( 2013 ) .
s. d. bass , a. w. thomas , _ acta phys .
* b45 * ( 2014 ) in print , arxiv:1311.7248 [ hep - ph ] .
s. hirenzaki _ et al .
, acta phys . pol .
_ * b41 * , 2211 ( 2010 ) .
h. nagahiro , s. hirenzaki , _ phys .
lett . _ * 94 * , 232503 ( 2005 ) .
h. nagahiro _ et al .
lett . _ * b709 * , 87 ( 2012 ) .
e. friedman , a. gal , j. mares , _ phys .
lett . _ * b725 * , 334 ( 2013 ) .
a. cieply _ et al .
_ , arxiv:1312.1547 .
s. wycech , w. krzemien , _ acta phys .
_ * b45 * ( 2014 ) in print , arxiv:1401.0747 [ nucl - th ] .
a. m. green , s. wycech , _ phys .
rev . _ * c71 * , 014001 ( 2005 ) .
n. g. kelkar _ et al .
phys . _ * 76 * , 066301 ( 2013 ) .
bass _ et al .
, acta phys . pol .
_ * b41 * , 2239 ( 2010 ) .
h. nagahiro _ et al . ,
rev . _ * c87 * , 045201 ( 2013 ) . s. klimt _ et al .
phys . _ * a516 * , 429 ( 1990 ) .
d. jido _ et al .
_ * a914 * 344 ( 2013 ) .
h. nagahiro , m. takizawa , s. hirenzaki , _ phys .
* c74 * , 045203 ( 2006 ) .
s. brauksiepe _ et al .
instrum . meth .
* a376 * , 397 ( 1996 ) .
p. klaja _ et al .
, aip conf.proc._ * 796 * , 160 ( 2005 ) .
moskal _ et al .
lett . _ * 80 * , 3202 ( 1998 ) .
moskal _ et al .
lett . _ * b474 * , 416 ( 2000 ) .
a. khoukaz _ et al .
j. _ * a20 * , 345 ( 2004 ) .
klaja _ et al .
* b684 * , 11 ( 2010 ) .
e. czerwiski , phd dissertation , jagiellonian university ( 2009 ) ; arxiv:0909.2781 .
e. czerwiski , p. moskal _ et al .
lett . _ * 105 * , 122001 ( 2010 ) .
p. moskal _
et al . , phys .
lett . _ * b482 * , 356 ( 2000 ) .
r. maier _ et al .
instrum . meth .
* a390 * , 1 ( 1997 ) .
p. moskal , _ et al .
* a466 * , 448 ( 2001 ) .
d. albers _ et al .
phys . j. _ * a22 * , 125 ( 2004 ) .
d. albers _ et al .
lett . _ * 78 * , 1652 ( 1997 ) .
f. hibou _ et .
al . , phys .
_ * b438 * , 41 ( 1998 ) .
f. balestra _ et .
al . , phys .
_ * b491 * , 29 ( 2000 ) .
g. fldt and c. wilkin , _ phys .
lett . _ * b382 * , 209 ( 1996 ) .
g. fldt and c. wilkin , _ phys .
rev . _ * c56 * , 2067 ( 1997 ) .
|
we describe a new high precision measurement of the production cross - section for the @xmath0 meson in proton - proton collisions @xmath1 for five beam momenta at low access energy region @xmath2 conducted at the detection system together with an updated results of all other previous measurements of @xmath1 at .
|
the generalized second law of thermodynamics@xcite is the statement that the entropy outside event horizons plus the bekenstein - hawking entropy @xmath1 ( in planck units ) of all event horizons can not decrease .
the law seems to be correct , at least in quasistationary processes@xcite .
if it is true , it must be that @xmath1 is the most entropy that could possibly be contained in a region bounded by an area @xmath2@xcite .
there has been much debate over the past 20 years about whether or not this bound really holds , and part of the problem in proving it is that it is not precisely clear what the statement means .
nevertheless , there are many reasonable arguments in support of it .
if the @xmath1 bound is indeed valid , then the the number of states is vastly overestimated by any ordinary flat space density of states in three dimensions .
the reason is that almost all of those states would , because of gravity , create a black hole whose entropy is fixed at @xmath1 .
one is thus led to the _ holographic hypothesis_@xcite ,
according to which the true degrees of freedom are enumerated on a surface enclosing the volume of interest , at an information density less than or equal to one bit per planck area .
susskind proposed to implement t hooft s holographic hypothesis by mapping all the points of space , by light rays that impinge perpendicularly , onto a flat two - dimensional _ screen _ in a distant asymptotically flat region .
we call this mapping the _ screen map_. this particular idea was partly motivated by properties of string theory in the light cone gauge , but the mapping between surface and volume degrees of freedom is not really specified in susskind s proposal
. the proposal seems more intended to get some kind of picture on the table so one can begin thinking about it . the first test to which susskind subjected the screen map was to ask whether the horizons of any black holes that are present are necessarily mapped onto sets of _ larger area _ on the screen
this is required by the holographic hypothesis since the black hole has the maximal bit density of one per unit area .
we call a screen map with this property an _ expanding map_. susskind argued that his screen map is indeed expanding .
he also noted that it is a one to many map from horizon to screen , and suggested that at the quantum level one should perhaps superpose all of the images . in this paper we take a closer look at the definition and properties of the screen map .
we find that , for a single screen time , there are an infinite number of multiple images , corresponding to _ different _ horizon times . since they arise from different time slices of the horizon , it does not seem appropriate to superpose them
. moreover , almost all of these single image maps are _ contracting _ rather than expanding , which is allowed by the focusing equation on account of the presence of focal points .
however , we prove in general and illustrate in several examples that there is always at least one expanding image of the horizon .
this proof is essentially hawking s area theorem , applied to the boundary of the past ( or future ) of the screen .
we are interested in light rays that hit the screen orthogonally at one screen time , that is , at one particular spacelike slice of the screen s history .
the first thing to clarify is whether the light rays leave the screen towards the future or towards the past . lacking the holographic theory
, we do not have a way to decide this , so we shall consider both possibilities .
if the rays from the screen are _ past directed _ , then they never actually cross the future black hole horizon .
they can however cross the
stretched horizon " @xcite , which in any case might be argued to be more relevant than the true event horizon .
if the rays from the screen are _ future directed _ , then they can indeed cross the future black hole horizon , so we get a precise map from a subset of to the horizon by following the rays .
we define the _ future screen map _ to be the past directed congruence of null rays orthogonal to . a point on a given ray
is mapped to the point where the ray hits the screen .
similarly we define the _ past screen map _ reversing the roles of past and future . in a static spacetime , such as the maximally extended schwarzschild metric , these are equivalent . in particular , although the future screen map does not intersect the future horizon , it does intersect the past horizon . in such cases
the past horizon can serve as a surrogate for the stretched horizon .
of particular interest will be the part of this congruence that lies on the _ boundary _
@xmath3 $ ] of the past of .
we refer to this part as the _ primary _ future screen map , and similarly for the past screen map .
the definition of the past screen map is illustrated in fig .
[ screenmap ] for the case in which a single black hole is present .
the intersection of the screen map with the horizon is shown .
the multiple coverings of the horizon come from rays that orbit the hole before crossing the horizon .
only the first covering lies in the primary screen map . rather than taking the screen to be planar
, it might be natural to think of it as a large sphere in an asymptotically flat spacetime , which can be taken all the way out to future or past null infinity .
the distinction between planar and spherical screens makes no difference for our general arguments ( as long as the rays orthogonal to the screen are not diverging ) , but the specific examples considered below will refer to a planar screen at infinity . in @xcite it was argued that the screen map is necessarily expanding as a consequence of the _ focusing equation _ @xmath4 which governs the rate of change with respect to affine parameter of the convergence @xmath5 of a congruence of null geodesics .
@xmath6 is the shear tensor , @xmath7 is the ricci tensor , and @xmath8 is the tangent vector to the congruence .
according to the einstein equation one has @xmath9 , so the rate of change @xmath10 is positive as long as the null energy condition @xmath11 holds .
thus , assuming the null energy condition , one knows that if the light rays have vanishing convergence at the screen , @xmath5 must be non - negative everywhere from screen to horizon ( with affine parameter increasing from screen to horizon ) , so the screen map is expanding . there is a serious flaw in this reasoning however since @xmath5 may become infinite somewhere , after which point it can be _
negative_. the focusing equation still says it must increase after that , but that is of no help in establishing the expanding character of the map .
a point where @xmath5 is infinite is a _ focal point _ ( also often called a caustic " or a conjugate point " to the screen ) . to be sure of the expanding character of the map
one must show that there are no focal points between the screen and the horizon .
the _ boundary _ of the past of the screen @xmath3 $ ] is what we have called the primary screen map .
this boundary is similar to a black hole event horizon , and shares with such a horizon the property that the area of its cross sections can not decrease toward the future@xcite .
this property follows from the focusing equation , the null energy condition , and a key property of all past boundaries : each point lies on a null geodesic that runs all the way up the boundary to with no focal points along the way .
the proof assumes that no naked singularities " , i.e. singularities visible from the screen , are encountered along the way up to .
thus one has an area theorem " for the primary screen map , which guarantees that this map is expanding .
in particular , if the null rays cross a stretched horizon , then the stretched horizon will necessarily be mapped to a larger area on the screen .
all the above applies equally well to the boundary of a past screen map , with the roles of future and past interchanged .
the boundary has an area that must _ decrease _ toward the future , but still _ increases _ toward the screen , so again one obtains an expanding map to the screen ( assuming now that no singularities are encountered on the way from the screen ) .
in this section we look at a number of specific examples that illustrate the general principles already discussed . in some of the examples we can actually calculate the areas of the screen images of the black hole horizon and show that a finite number ( often only one ) are greater but the remaining infinite number are less than the horizon area .
we explain this fact by identifying the focal points .
we also identify the primary screen map in all the examples and show explicitly that it has no focal points . since these are static spacetimes the future and past screen maps are equivalent .
for definiteness , we adopt here the past screen map , which intersects the future black hole horizon . for a single spherically symmetric black hole we can easily analyze the screen map explicitly by use of the integrated geodesic equation .
three orbits from the screen to a schwarzschild hole are shown in figure [ threeorbits ] .
the orbits can be labeled by the angle at which they hit the horizon , measured from the perpendicular from the screen to the hole .
the orbits with angles between 0 and @xmath12 cover the horizon once , when rotated around the axis . those with angles between @xmath12 and @xmath13 give a second covering , and so on .
there are an infinite number of coverings of the horizon .
there is an upper bound for the impact parameter for capture by the hole , so the areas of the coverings on the screen must converge to zero .
these coverings are illustrated in figure [ bullseye ] , which is drawn approximately to scale for the case of a schwarzschild black hole .
the primary cover has area @xmath14 , greater than the horizon area , but all the rest have lesser area .
in fact the sum of all the rest has lesser area , @xmath15 . since the higher order covers are not expanding , there must be at least one focal point somewhere on each of the corresponding null congruences .
this focal point is shown in figure [ focalpt ] for the second cover in the schwarzschild case .
all the null rays will intersect at a point directly behind the black hole ( as viewed from the screen ) . at this point
the convergence goes to infinity and just past this point it is negative ( although in general the convergence may become positive again and may even diverge again if the congruence goes through another focal point ) .
it is interesting to note that the separation of the null rays in the transverse direction is larger on the screen than on the horizon , so it is not the angular focusing but the radial focusing that makes the map fail to be expanding , even though it is the angular focusing that produces the focal point .
we also analyzed the screen map for an extremally charged reissner - nordstrom hole .
there it turns out that both the primary and the second cover are expanding , but the rest are not .
it was pointed out in @xcite that one black hole can not be hidden behind another , and that the screen map will still be expanding for both horizons in this case .
we investigated this situation explicitly for the case where the two black holes are very far from one another and the screen is perpendicular to the axis joining them . in this case
we could approximate the capture orbits for the second hole analytically by simply composing with the scattering by the first hole .
the screen map pattern in this case is rather complicated .
a general orbit can alternate between the two holes , wrapping any number of times around on each visit , before finally crossing the horizon of one of the holes .
the primary map is given by the orbits that never cross the axis ( see figure [ 2holes ] ) , and it traces out an annulus on the screen .
we estimated the area of this annulus for large separation @xmath16 , and found that it grows like @xmath17 : @xmath18 . here
@xmath19 , @xmath20 are the masses of the first and second holes respectively , @xmath21 is the horizon area of the second , and we have assumed that @xmath22 .
for two nearby black holes one can not just compose the asymptotic scatterings , and also one must decide what metric to take for the two black holes .
in order to have a static situation , it is natural to consider a pair of extremally charged black holes with like sign charges , one of the majumdar - papapetrou solutions@xcite .
generalizing this , one can consider a linear array of any number of such black holes .
we shall now show by explicit consideration of the geodesics that the primary map has no focal points , so the map is expanding .
it is most natural to discuss this problem for an arbitrary static axisymmetric geometry with the screen perpendicular to the axis .
because of the axisymmetry , focal points will occur on any rays that cross the axis , so we restrict attention to those rays that do not cross the axis . these can only have focal points if they are focussed to crossing points in the radial direction . to establish the absence of such crossing points
we adopt a technique from the analysis by yurtsever@xcite of chaos in the orbits around a pair of extremal black holes .
consider a line element of the form @xmath23 , where @xmath24 , and @xmath25 and @xmath26 are functions only of @xmath27 .
the rays from the screen have constant @xmath28 , and , since they are null geodesics , they are the same as for the conformally rescaled three - dimensional line element @xmath29 , where @xmath30 .
the geodesics of @xmath31 project to geodesics of the riemannian metric @xmath32 , and the arc length along the projection agrees with the affine parameter along the null geodesic .
now computation reveals that the curvature of the two - dimensional metric @xmath32 is negative everywhere outside the horizon of a general multi extremal black hole solution , as well as for a reissner - nordstrom black hole with any charge to mass ratio .
thus in all these cases the projected curves are receding from each other .
the spacetime null geodesics are therefore also receding with respect to @xmath31 . while this does not imply that they are receding with respect to the physical metric @xmath33 ,
it does at least imply that they will not reach a focal point .
the main lesson of this geometric exercise is the conclusion that , to guarantee the expanding nature of the holographic map from horizon to screen , one should restrict to the _ primary _ screen map , i.e. the boundary of the past ( or future ) of the screen .
this boundary generalizes the concept of a black hole event horizon and satisfies the area theorem .
even the primary screen map is only guaranteed to be expanding if the null energy condition is satisfied . in the presence of quantum fields this condition
can be locally violated by the expectation value of the stress - energy tensor , so one should not rely on the null energy condition .
perhaps an _ averaged _ null energy condition would be sufficient to establish the expanding nature of the primary screen map .
( several recent works@xcite have studied the extent to which averaged energy conditions hold in quantum field theory . ) alternatively , the expanding property may be lost , which may reveal something about the holographic principle .
our proof of the expanding nature of the primary screen map relied heavily on the assumption that the rays orthogonal to the screen are not diverging .
this is guaranteed in asymptotically flat space by taking a flat or spherical screen at infinity . in a closed expanding universe
it seems one must choose a future screen map rather than a past one in order to have any chance of finding an expanding map from all points of space to a single screen .
strangely , if the universe recollapses , one must then switch to a past screen map .
perhaps some insight into the holographic hypothesis can be gleaned from an investigation of its compatibility with closed universes .
this work was supported in part by nsf grant phy-94 - 13253 , a umd graduate fellowship , and the university of utrecht .
w. zurek , _ phys .
* 49 * ( 1982 ) 1683 ; w.h .
zurek and k.s .
thorne , _ phys .
* 54 * ( 1985 ) 2171 ; k.s .
thorne , w.h .
zurek and r.h . price , in _ black holes : the membrane paradigm _ ,
eds . k.s .
thorne , r.h .
price and d.a .
macdonald ( yale university press , new haven , 1986 ) p. 280 ; w.g .
unruh and r.m .
wald , phys .
d * 25 * , 942 ( 1982 ) ; phys .
d * 27 * , 2271 ( 1983 ) ; correction : m.j .
radzikowski and w.g .
unruh , phys .
d * 37 * , 3059 ( 1988 ) ; r.m .
wald , in _ quantum field theory in curved spacetime and black hole thermodynamics _ ,
( university of chicago press , 1994 ) ; v.p .
frolov and d.n .
page , _ phys .
* 71 * ( 1993 ) 3902 .
g. t hooft , dimensional reduction in quantum gravity " , in _ salamfestschrift : a collection of talks _ , world scientific series in 20th century physics , vol .
a. ali , j. ellis , s. randjbar - daemi ( world scientific , 1993 ) .
r.m . wald and u. yurtsever ,
d * 44 * ( 1991)403 ; a. folacci , phys
d * 46 * ( 1992 ) 2726 ; m. visser , phys .
b349 * ( 1995 ) 443 ; u. yurtsever , phys .
d * 51 * ( 1995 ) 5797 ; l.h .
ford and t.a .
roman , phys .
d * 51 * ( 1995 ) 4277 ; u. yurtsever , phys .
d. * 52 * ( 1995 ) 564 ; l.h .
ford and t.a .
roman , averaged energy conditions and evaporating black holes " , mit - ctp-2446 , gr - qc/9506052 .
|
the screen mapping " introduced by susskind to implement t hooft s holographic hypothesis is studied . for a single screen time , there are an infinite number of images of a black hole event horizon , almost all of which have smaller area on the screen than the horizon area .
this is consistent with the focusing equation because of the existence of focal points .
however , the _ boundary _ of the past ( or future ) of the screen obeys the area theorem , and so always gives an expanding map to the screen , as required by the holographic hypothesis .
these considerations are illustrated with several axisymmetric static black hole spacetimes .
@xmath0 utrecht - thu-96/11 + gr - qc/9602043 + * focusing and the holographic hypothesis * + .5 cm _
institute for theoretical physics , university of utrecht + p.o .
box 80.006 , 3508 ta utrecht , the netherlands _ + and + _ department of physics , university of maryland + college park , md 20742 - 4111 , usa _
+ corley , [email protected]
|
the aurivillius phases are layered bismuth compounds that obey the general formula bi@xmath4a@xmath5b@xmath6o@xmath7 @xcite .
the family includes many members that present ferroelectricity at room temperature , and they have been widely studied for potential technological applications mainly in thin film nonvolatile memories.@xcite srbi@xmath0ta@xmath0o@xmath1 ( sbt ) and srbi@xmath0nb@xmath0o@xmath1 ( sbn ) are members of the family , with @xmath8 and @xmath9 and their structure is formed by alternating two srmo@xmath10 ( m@xmath11ta , nb ) perovskite blocks and one bi@xmath0o@xmath0 slab ( fig .
[ fig : structure ] ) .
tantalum and niobium present very similar physical and chemical properties , including the valence and atomic radii . as a result , sbt and sbn present isomorphous polar structures at room temperature @xcite and analogous mechanical and electrical characteristics .
however , their phase diagrams are qualitatively different : on increasing the temperature sbt undergoes a phase transition to a non - polar orthorhombic phase that does not arise in sbn .
@xcite previous first - principles calculations by perez - mato _ et al _ revealed that the trilinear coupling of two primary unstable modes and a secondary hard mode is critical to stabilize the ground state of sbt .
@xcite this work provided a simple and plausible scheme in contrast with others alternatives where the eventual existance of negative biquadratic coupling triggers the simultaneous condensation of several modes .
@xcite later , the key role of the trilinear coupling has been found in several other compounds@xcite and its influence has been proposed as the origen of the so - called _ _
avalanche__@xcite phase transitions and the _ hybrid _ improper ferroelectricity.@xcite in this work we revisit the sbt and extend the _ ab initio _ study to sbn in order to investigate the role of the trilinear coupling in the stabilization of the ferroelectric structure and its influence in the sequences of phase diagrams .
the high- and low - temperatures structures of sbt and sbn are isomorphous : tetragonal with space group @xmath12 ( no .
139 ) at high enough temperatures@xcite and orthorhombic @xmath13 ( no . 36 ) at low temperatures . in the case of sbn
there is a direct phase transition from the tetragonal to the ferroelectric structure at 800k,@xcite whereas sbt presents an intermediate nonpolar orthorhombic @xmath14 phase ( no . 63 ) between 500 - 600k and 770k .
the cell parameters of the two orthorhombic phases are related to the tetragonal ones according to @xmath15 , @xmath16 and @xmath17 , which means that the volume of the unit cell is doubled in the tetragonal - to - orthorhombic transitions .
the tetragonal space group ( @xmath12 ) is a supergroup of the two orthorhombic space groups , the intermediate phase of sbt ( @xmath14 ) and the ground state of both compounds ( @xmath13 ) . in consequence
, the low temperature structures can be described in terms of symmetry - adapted distortions with respect to the parent tetragonal structure .
@xcite table [ tab : mode_decomposition ] shows this decomposition in terms of the symmetry - adapted modes that break the tetragonal symmetry for the experimental structures of sbt and sbn obtained with the aid of the tool amplimodes ( www.cryst.ehu.es ) .
@xcite the hierarchy of the modes is similar in both compounds : the amplitude of @xmath18 is dominant and can be considered as the primary order parameter and the @xmath19 distortion seems to play a secondary role .
the analysis of the @xmath20 polar distortion is more delicate because due to the polar character of the ground state its amplitude depends on the origin . in table
[ tab : mode_decomposition ] the chosen origin does not change the arithmetic center of the parent structure . according to the isotropy subgroups the intermediate phase of sbt is naturally explained as a condensation of the @xmath18 mode , and in terms of the landau theory of phase transitions the @xmath21 transition can be continuous .
the second transition to the ferroelectric structure should then consist on the condensation of the polar mode . in the case of sbn a direct transition between the tetragonal and the ferroelectric phase
can be solely described by a simultaneous condensation of at least any pair of the three modes , and according to landau theory , it should be first order : the three relevant modes condense simultaneously in an _ avalanche _ phase transition @xcite .
the orthorhombic strain is usually very small in the aurivillius compounds and sbt is not an exception .
the experimental values of cell parameters in sbt are @xmath22 and @xmath23 for the tetragonal structure , @xcite and @xmath24 , @xmath25 and @xmath26 for the ferroelectric phase.@xcite the deformation of the cell is very small with @xmath27 ( all within a 0.5 @xmath28 margin ) , and the influence of strain can be considered as a residual effect . up to our knowledge , there is not such an experimental data available for the tetragonal cell of sbn , and the orthorhombic cell parameters are @xmath29 and @xmath30 and @xmath31 .
@xcite a complete description of the atomic displacements associated with the three modes can be found in ref . .
in summary , among the 18 degrees of freedom that constitute a generic distortion that links the tetragonal and ferroelectric structure , seven correspond to @xmath32 , and they mainly account for tiltings of the oxygen octahedra around the @xmath33 direction of the tetragonal cell and displacements of the bi cations along the @xmath34 direction .
the octahedra behave as rigid units and , in consequence , neighboring octahedra show antiphase tiltings .
the polar @xmath35 distortion involves essentially an antiphase displacement of the bi atoms and the perovskite blocks , along the @xmath33 direction . among the eight independent displacements associated with the @xmath36 distortion
the main amplitudes correspond to tiltings of the oxygen octahedra around the @xmath37 axis and antiphase displacements of the oxygens in the bi@xmath0o@xmath0 slabs along the @xmath33 direction .
.summary of the mode decomposition of the experimental and calculated distortions of the ferroelectric phases of sbt and sbn with respect to their corresponding parent ( calculated and experimental ) structures .
amplitudes are given for modes normalized within the primitive unit cell of the polar structure .
the second row shows in parentheses the scalar product of the calculated and experimental polarization vectors after normalization . [ cols="^,^,^,^,^,^,^,^ " , ]
despite the different sequence of phase transitions that sbt and sbn go through , first principles calculations shows that the nature and main features of the instabilities in both compounds are qualitatively very similar , and that most of the conclusions about sbt obtained in ref . can be extended to sbn . among the three symmetry adapted distortions ( @xmath32 , @xmath35 and @xmath36 ) that take part to drive the tetragonal @xmath12 parent phase to the polar @xmath13 structure
, @xmath32 is dominant with the highest amplitude and the deepest energy well ( @xmath38 11 mry per formula unit ) .
the depths of the secondary distortions @xmath35 and @xmath36 are similar in sbn ( @xmath38 3.5 mry p.f.u . ) , while in sbt @xmath36 is hardly unstable and @xmath35 slightly deeper ( 5.2 mry p.f.u . ) . in both compounds , the simultaneous condensation of the primary instability @xmath32 with any of the other two distortions
is penalized energetically by positive strong biquadratic couplings .
thus , the trilinear coupling is the key ingredient that allows the simultaneous condensation of the three modes and its role is essential to stabilize the observed ground state .
the most noticeable difference between sbt and sbn is associated with the magnitude of the trilinear term .
the polynomial expansion of the energy of both compounds shows that the trilinear coupling is much stronger in sbn than in sbt , suggesting that its magnitude could be crucial to suppress the intermediate phase in sbn .
moreover , the analysis of phenomenological phase diagrams for sbt and sbn shows that a higher value of this constant is enough to change the topology of the phase diagram , allowing a direct phase transition from the parent phase to the ground state . finally , a simplified @xmath3 hamiltonian that retains the symmetry requirements and the correct dimensionality of the order parameters has been developed for the sbt / sbn system .
monte carlo simulations do not show qualitatively differences in comparison with the one - dimensional case@xcite : the different sequence of phase transitions observed in both compounds can be reproduced by changing the strength of the trilinear coupling .
however , according to the present calculations , the increment of the fluctuations associated with the higher dimensionality of the order parameters tends to favour the suppression of the intermediate phase and to reinforce the first order character of the direct transition between the high- and low - temperature structures .
this work has been supported by the basque government ( project no .
it-779 - 13 ) and the spanish ministry of education and science ( project no .
mat2012 - 34740 ) .
one of the authors ( u.p . ) thankfully acknowledges the financial support of the university of the basque country ( upv / ehu ) .
|
despite their structural similarities , srbi@xmath0ta@xmath0o@xmath1 ( sbt ) and srbi@xmath0nb@xmath0o@xmath1 ( sbn ) undergo a different sequence of phase transitions .
the phase diagram of sbt as a function of the temperature includes an intermediate phase between the high - temperature phase and the ferroelectric ground state , while in the niobium compound the intermediate phase is suppressed and a single transition between the high- and low temperature structures is observed .
we present @xmath2 calculations that reveal the relevance of a trilinear coupling between three symmetry - adapted modes to stabilize the ground sate in both compounds , being this coupling much stronger in sbn . within the framework of the phenomenological landau theory ,
it is shown that by solely increasing the strength of the trilinear coupling the topology of the phase diagram of sbt can change up to suppress the intermediate phase .
monte carlo simulations on an idealized @xmath3 hamiltonian confirm that the trilinear coupling is the key parameter that determines the sequence of phase transitions and that for higher dimensionality of the order parameters the stability region of the intermediate phase is narrower .
|
disk accretion is attributed to cold cosmic flows @xcite and dwarf satellite capture @xcite through the virial radius , which may be 20 times larger than the disk ( see reviews in * ? ? ?
* ; * ? ? ?
it is not clear whether these flows reach the disk directly . @xcite and @xcite suggest that the magellanic stream breaks apart and gets ionized by hydrodynamic instabilities without falling onto the disk .
@xcite also found substantial heating and ionization of the layers surrounding a cosmic cold flow in simulations of the milky way .
the numerical algorithm used for the simulation might also be important ; the arepo code @xcite has such high resolution that the infalling gas breaks apart by instabilities and heats up before reaching the disk .
these predictions are consistent with the observation that the inflow rate of neutral clouds around the milky way is small , @xmath0 yr@xmath1 @xcite , which is too low to maintain the star formation rate of several @xmath2 yr@xmath1 @xcite if that is the only gas that comes in .
low accretion leads to a relatively rapid quenching of star formation , contradicting the observation that most disk galaxies form stars over a high fraction of the hubble time ( e.g. , * ? ? ?
cosmic simulations suggest that the accreting gas moves toward larger radii over time as the specific angular momentum of the remaining cosmic filament increases ; this leads to the popular picture of inside - out star formation @xcite .
observations agree that accretion occurs today in the outer parts of galaxies @xcite .
this causes a problem if most of the star formation takes place in the main disk , much further inward .
why does the star formation rate equal the accretion rate , as analytical models suggest @xcite , if these rates occur in different locations ? one needs in - plane accretion at a comparable rate to deliver the fresh outer gas to the inner star - forming regions .
@xcite discuss how this might happen using a model based on energy dissipation .
@xcite consider in - plane accretion that is driven by disk torques from interacting clumps and turbulent viscosity .
another model of disk accretion assumes that cosmic gas goes first into a hot halo and then condenses onto the disk in a second step via thermal cooling at the interface .
a galactic fountain could provide the cool surface needed for this condensation @xcite .
the fountain model avoids the question of whether the cosmic gas goes directly from the virial radius to the disk , and it avoids the problem of accretion onto the outer regions when the gas is needed in the inner regions to form stars .
cosmic gas may be deposited in the far outer regions by cold streams while the inner disk is fed by condensing halo gas immediately adjacent to it .
@xcite summarize the advantages of this model , including many comparisons to observations @xcite .
halo condensation gives an accretion rate that is approximately proportional to the star formation rate at the same position , and therefore to the existing column density of molecular or self - gravitating gas .
such a model sustains star formation in the main disk .
cosmic accretion should generally have a different specific angular momentum than the part of the disk that receives it .
there will be some generation of turbulence to dissipate the relative kinetic energy , and an inflow or outflow of the combined gas until radial force equilibrium is achieved in a new circular orbit . even in the fountain model , where the accreting gas in the disk comes directly from the adjacent rotating halo , the high velocity dispersion in the halo and the resulting radial pressure gradient
cause it to rotate more slowly than the cold disk beneath it .
the result will be a gradual shrinkage of the gas disk as it forms stars from condensing halo gas .
this paper calculates the radial drift speed from halo condensation at the local star formation rate in order to determine how such a drift affects the radial profiles of the gaseous and stellar components of the disk .
the results suggest a relatively long timescale for disk evolution forced this way , so the process may not have strong observational signatures in disks dominated by other torques , such as bars and spirals .
however it may be evident in dwarf galaxies without these stellar structures .
dwarf galaxy star formation also seems to proceed from the outside - in @xcite , as if following some inflowing gas , unlike the case in spiral galaxies .
the star formation history in the large magellanic cloud shows an outside - in trend too , although interactions with the small magellanic cloud could be involved @xcite .
other processes involving extra - planar gas should stimulate the same type of halo condensation and disk shrinkage .
@xcite suggested that gas at the disk - halo interface follows the motions of spirals and bars , which concentrate star formation and increase the turbulent pressure locally .
they model the resulting fountain and explain the vertical rotational velocity gradient that is observed in the interface region .
@xcite simulated the generation of turbulence by instabilities at high latitude above spiral density wave shocks ; turbulent structures like this could enhance halo condensation too . in what follows ,
the equations for disk evolution driven by accretion of low angular momentum gas are solved for the case where the accretion rate is proportional to the local star formation rate .
solutions are shown that include depletion of gas by star formation and disk winds , and with rates for accretion and star formation that change over cosmological time .
we show that the star - forming part of the gas disk shrinks in a self - similar way throughout this process , which means that an initially exponential profile for this component stays exponential with a shrinking scale length .
this is important because exponential profiles appear very early in the universe @xcite , perhaps as a result of clumpy disk dynamics @xcite , and subsequent accretion at the star formation rate should not distort this profile too much .
specific applications to blue compact dwarfs are discussed in section [ bcdsection ] .
we assume a that galactic disk accretes locally at a rate equal to @xmath3 in units of mass per area per time . in the fountain model
, this accretion is affected primarily by the star - forming component of the disk , so we divide the gas column density into two components , @xmath4 for the star forming part , such as the molecular and self - gravitating clouds , and @xmath5 for the non - star - forming part , such as the atomic diffuse gas . these two components interchange mass and mix with accreted gas on dynamical timescales , but only the fraction of the total that is involved with star - formation will drive the accretion and get depleted as stars form . with this assumption ,
the star formation rate is @xmath6 , with a fraction @xmath7 of the forming stellar mass quickly returned to the gas in the form of supernovae and stellar winds . in general
, there is also a disk wind .
a common approximation is to use a mass loss rate in the wind proportional to the star formation rate , @xmath8 , for constant @xmath9 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
then the continuity equation for gas surface density , including radial drift speeds from angular momentum redistribution , @xmath10 and @xmath11 for the star - forming and non - star - forming parts , is @xmath12 we also assume that cosmic accretion has some rotation speed , @xmath13 , that is not generally the same as the circular speed of gas that is already in the disk at that radius , @xmath14 .
the equation for conservation of angular momentum for the two components is then @xmath15 the left - hand side of this equation is the time derivative of the local disk angular momentum , following the radial drift , and the right - hand side is the angular momentum added to this part of the disk . assuming the circular speed changes only slowly with time , we separate @xmath16 from the first term on the left of equation ( [ ang ] ) and divide the whole equation by @xmath17 .
a term like @xmath18 enters so we assume a power law rotation curve of the form @xmath19 , which makes this term equal to @xmath20 . with these modifications to equation ( [ ang ] )
we subtract equation ( [ cont2 ] ) to eliminate @xmath21 , several common geometric terms , and the wind and star formation terms .
the result is @xmath22 for the star - forming component .
we also get @xmath23 for the non - star forming component , leaving @xmath5 unchanged , because it has no source terms on the right - hand sides of equations ( [ cont2 ] ) and ( [ ang ] ) .
this result is consistent with the finding by @xcite , @xcite and others that the atomic phase of interstellar gas has approximately constant mass over time with the accretion feeding only the star formation component .
our modification here is to consider that the accreted gas has a lower specific angular momentum than the disk gas , in which case the fraction of the disk mass involved with star formation moves inward to conserve total angular momentum ( eq .
[ vrequation ] ) .
this result is also consistent with the observation that the radial distribution of hi is usually much flatter than the molecular distribution , which matches that of the star formation rate and approximately also that of the underlying disk stars .
we assume that @xmath24 is proportional to the column density of disk gas that is involved with star formation , @xmath25 because of the galactic fountain process described in the introduction , along with a kennicutt - schmidt relation that is linear between the star formation rate and the dense component of the gas column density @xcite . in the model by @xcite and collaborators ,
the accreted gas is from a large reservoir in the hot halo , in which case it was formerly in hydrostatic equilibrium in the potential well of the galaxy and did not have much rotational support .
observations indicate a @xmath26%-20% slower rotation of the disk - halo interface than the disk itself ( e.g. , * ? ? ?
this disk - halo interface is where the condensation of halo gas onto the fountain is presumably happening , so the rotation speed there is already a mixture of the disk speed from supernova and wind ejecta , and the halo speed from hot gas condensing onto it .
the unseen halo speed is probably even lower than the observed interface speed . for simplicity
, we write , @xmath27 in what follows , this ratio will only be compared to unity , so slightly different radial variations in @xmath13 and @xmath14 will not matter much . the solution to equation ( [ vrequation ] ) is now @xmath28 we interpret this equation to mean that the star - forming part of the gas moves in at the speed @xmath10 .
this does not mean that individual molecular clouds are drifting at that speed .
as mentioned above , there should be a continuous mixing between the star - forming and diffuse parts following star formation , stellar gas return , molecular cloud disruption , and so on .
thus , equation ( [ vrsol ] ) may also be interpreted to mean there is an inward drift of the whole interstellar medium at the speed @xmath29 , driven by the star - forming part through its interaction with the slower - moving halo .
if accretion enables star formation at about the same rate , and star formation drives accretion as in the fountain model , then @xmath30 , giving @xmath31 .
thus , we can equivalently interpret our equations to mean that the star formation rate scales linearly with the fraction @xmath32 of the total gas , and this fraction drifts at @xmath10 .
the linear dependence of @xmath10 on @xmath33 implies a self similar shrinking . to solve for the resulting @xmath34 , we substitute @xmath10 into equation ( [ cont2 ] ) and simplify to get @xmath35 in the discussion above
, the entire coefficient of the @xmath4 term on the right is independent of radius but not necessarily time , so we represent it by @xmath36 then we have an evolution equation of the form @xmath37 for dot and prime representing time and radial derivatives .
the first term on the right can be eliminated by substituting @xmath38 .
then @xmath39 this equation is solved by setting @xmath40 equal to any function @xmath41 of the variable @xmath42 , because then @xmath43 and @xmath44 .
thus the star - forming part of the gas column density becomes @xmath45 @xmath46 represents the radial profile at @xmath47 .
we consider an initial profile that is exponential , so @xmath48 for @xmath49 and initial scale length @xmath50 .
then @xmath51 with exponentially shrinking scale length , @xmath52 , and initial central value @xmath53 .
evidently , the accretion makes the central column density of this disk component increase with time as @xmath54 .
because we have assumed @xmath55 to be independent of radius , the radial profile of the star - forming part of the disk is always exponential following an initial exponential shape , even as it shrinks with the accretion of low - angular momentum halo gas .
we might expect @xmath55 to have some radial variation , however , even if the star formation rate is locally proportional to the accretion rate ( @xmath56 is constant ) , because of the likely radial variations of the rotation curve slope , @xmath20 , and the wind mass loading factor , @xmath9 ( cf . eqn .
[ cala ] ) .
the rotation curve should flatten with increasing @xmath33 , making @xmath20 decrease and @xmath57 increase .
the wind factor should increase with @xmath33 as the disk material is less bound to the galaxy in the outer parts .
these two effects compensate each other in @xmath58 .
the other parameters , namely , the gas return fraction @xmath7 and the differential circular speed , @xmath59 should not vary as much ( especially if @xmath60 ) . also , @xmath61 is not observed to vary with radius in local disk galaxies for the star forming part of the interstellar medium , i.e. , the co - emitting gas @xcite .
thus @xmath55 may not vary much .
the effects of variable @xmath55 can be understood intuitively because the inflow speed @xmath62 scales with this quantity .
locally high @xmath55 can make a trough , for example , by increasing the inflow rate through that region .
the star - forming gas mass in the disk is @xmath63 showing the same exponential growth .
this type of growth arises because the accretion rate is assumed to be proportional to the star - forming gas mass itself , by feedback through star formation , a galactic fountain , and condensation of halo gas onto the cool fountain debris .
the accretion rate will generally change with time because both @xmath64 and the rate of cosmic accretion into the virial radius change .
sample solutions derived numerically are shown in figures 1 and 2 for two models . on the left - hand side of each figure , we assume in the first model that the accretion rate equals the star formation rate , which gives @xmath65 .
following @xcite , we use @xmath66 , where @xmath67 is the hubble time . on the right - hand side of each figure , we assume in the second model that @xmath56 also scales with the halo accretion rate , normalized to the halo accretion rate at the present time : @xmath68 where @xcite @xmath69 @xmath70 is the mass of the halo in units of @xmath71 , and @xmath72 is the redshift , related to the expansion time @xmath67 by @xmath73cdm cosmology from @xcite where @xmath74 km s@xmath1 mpc@xmath1 and @xmath75 . for the evaluation of @xmath10 , we assume @xmath76 for all radii , so there is a 10% mismatch between the halo and disk spins .
we also assume a flat rotation curve , @xmath77 , in figure 1 , and a solid body rotation curve , @xmath78 , in figure 2 . from equation ( [ vrsol ] ) ,
we see that the evolution is slower for higher @xmath20 and lower @xmath79 .
the starting time for the accretion is taken to correspond to a redshift of @xmath80 , which is in the clumpy disk phase when the smooth exponential disk starts to appear ( e.g. , * ? ? ?
all of the surface density profiles are normalized in the figures to the central value of the initial profile .
the curves in the bottom panels show the instantaneous radial profiles at equally spaced times from the start ( green curve ) to today ( red curve ) .
the profiles remain exponential as they steepen , as implied by equation [ solution ] .
the top panels in the figures show the stellar surface densities calculated at each radius by integrating the local star formation rate over time , @xmath81 @xmath82 gyr at the start , where @xmath80 , and @xmath67 increases with cosmic time until today , when @xmath83 gyr . because @xmath4 is normalized to the central value at the start , the stellar surface density is not scaled to real units .
we assume @xmath84 here ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and for the gas evolution equation ( [ sigmadot ] ) , @xmath85 @xcite .
the stellar profiles steepen with time , but they do not remain exponential like the gas because the stars have the gas profiles from the times when they formed . also in the top panels , plotted with black curves and using the right - hand axes , are the average , mass - weighted ages of the stellar disks .
the average ages are smaller near the center because the gas has moved in and formed an increasing proportion of stars there over time .
this is outside - in star formation .
surface density evolution by fountain - driven accretion is faster for more massive galaxies because of the @xmath86 term in @xmath10 ; massive galaxies tend to have flatter rotation curves in their bright inner regions @xcite .
the age dip at the center is also sharper for higher mass galaxies , for the same reason .
because we have ignored bulge formation and other accretion processes that happened earlier , the present results are only meant to show the part of the evolution from fountain - driven accretion
. there could be an application to late - type galaxies .
blue compact dwarfs ( bcds ) have high ratios of gas mass to star mass , low metallicities , nearly solid - body rotation curves , and a generally young appearance .
they also have steeper and bluer exponential profiles in the inner disk than the outer disk @xcite .
dwarf irregulars tend to have bluer inner regions too , suggesting outside - in star formation @xcite .
for example , the far ultraviolet scale lengths for low mass galaxies are about half the 3.6@xmath87 m scale lengths , unlike the case for high - mass galaxies where these scale lengths are about the same @xcite .
low mass galaxies have weak or no spiral arms to drive a radial evolution , and they have sustained star formation at a steady , though bursty , rate all of their lives . if the freshly accreted gas that drives star formation has relatively low angular momentum , then there could have been a radial accretion like that discussed here .
figure 3 shows the radial profiles of stellar mass surface density in four dwarf galaxies with type iii stellar surface brightness profiles : exponential profiles with radius that have an abrupt break and become shallower in the outer disk ( see * ? ? ?
these profiles are common among bcds or dwarf galaxies with significant central star formation ( current or past ) .
haro 29 and ngc 3738 are classical bcds with young central parts and older , smooth outer parts @xcite .
ngc 1569 is a young starburst @xcite , and ddo 70 is a relatively normal dwarf irregular with evidence for an event in the past that produced an enhancement in the center of the galaxy @xcite .
each galaxy has steeper surface brightness and stellar mass surface density profiles in the inner disk compared to the outer disk , presumably because of some accretion that led to the starburst there .
the observed mass surface density profiles are shown as red curves ( from * ? ? ?
* ) , and compared to theoretical models of stellar surface density calculated as above .
the models fitted the inner and outer slopes primarily by adjusting the accretion history .
all models had the same values of @xmath88 ( a relatively large angular momentum mismatch between the halo and the disk ) , @xmath89 ( a slowly rising rotation curve ) , @xmath90 ( a relatively strong wind for these small galaxies , i.e. , @xmath91 as above but now @xmath92 for the wind ) , and a redshift of 2 at the start of the evolution , as above .
the two galaxies with the relatively small increase of slope in the center , ddo 70 and ngc 1569 , assume @xmath93 without the cosmic accretion multiplier in equation ( [ a2 ] ) , while the two galaxies with large increases of slope in the center , haro 29 and ngc 3738 , assume @xmath93 up to 1 gyr ago and then a burst of accretion toward the end with 10 times this rate . in the figure ,
the top blue curve is the one corresponding to the present time , while the other curves show the evolution in equal time steps .
these simple models fit haro 29 and ddo 70 very well ( haro 29 has a small centrally flat part for which we have artificially moved the galaxy profile slightly to the left in the figure ) . for ngc 1569 and ngc 3738
, the models fit the outer profiles well , but produce inner mass profiles that are slightly less massive than what are observed with roughly the same average slope .
thus , aside from some irregularities in the profiles , fits to the average inner and outer slopes are possible with the accretion model .
galaxies with steeper inner profiles have a larger recent influx of low angular momentum gas .
if the influx is steady , then the stars build up with a steadily decreasing scale length and there is a more gradual transition between the inner and outer parts . also shown in figure 3
are the gas surface density profiles derived from the deprojected hi mass surface densities multiplied by 1.36 to account for helium and heavy elements .
these profiles go out much further in radius than the stellar parts shown in the panels below each one .
they are plotted in physical units , @xmath2 pc@xmath94 , using the left - hand axes . the molecular gas surface density , which presumably follows the star formation rate and inner stellar disk better , is probably much lower than the hi+he surface density throughout most of the plotted region .
the hi+he profile is not exponential even though the profile of star formation rate is ( h@xmath95 profiles are in * ? ? ?
the interstellar matter in galactic gas disks should migrate inward if the disk accretes from the halo regions at slightly lower specific angular momentum .
accretion at the disk - halo interface @xcite could cause such migration and have a noticeable effect on the stellar age distribution after several gyr .
we have shown that this process preserves the exponential shape of the star - forming part of a gas disk initially in this form , while shrinking the disk scale length to conserve angular momentum .
the radial drift speed is small , on the order of the disk size divided by the hubble time , or @xmath96 km s@xmath1 .
gas consumption by star formation and gas expulsion by disk winds do not change the exponential shape nor the rate of evolution of the scale length ; they only change the absolute value of the surface density .
fits of the model to bcd galaxies can account for the steep inner stellar profiles of those galaxies while explaining the origin of the dense gas in the center that drives the starburst .
|
star formation in most galaxies requires cosmic gas accretion because the gas consumption time is short compared to the hubble time .
this accretion presumably comes from a combination of infalling satellite debris , cold flows , and condensation of hot halo gas at the cool disk interface , perhaps aided by a galactic fountain . in general
, the accretion will have a different specific angular momentum than the part of the disk that receives it , even if the gas comes from the nearby halo . then the gas disk expands or shrinks over time . here
we show that condensation of halo gas at a rate proportional to the star formation rate in the fountain model will preserve an initial shape , such as an exponential , with a shrinking scale length , leaving behind a stellar disk with a slightly steeper profile of younger stars near the center .
this process is slow for most galaxies , producing imperceptible radial speeds , and it may be dominated by other torques , but it could be important for blue compact dwarfs , which tend to have large , irregular gas reservoirs and steep blue profiles in their inner stellar disks .
|
defining urban form is certainly an important and difficult issue , especially if one wants to supply useful knowledge to urban planners and urban designers or new knowledge for city scientist . in this paper
we address this question , and we try to improve the empirical - based knowledge on the structure of a city by studying the urban street networks of ten european cities , namely edinburgh , leicester , sheffield , oxford , worcester , lancaster , catania , barcelona , bologna and geneva .
the form of cities is the subject of an area of urban studies named urban morphology .
urban morphology in its current form emerged between the 40s and the 60s of the xx century from the work of two scholars as prominent as different : the german - born and then british urban geographer m.r.conzen ( 1960 ) , and the italian architect and historian saverio muratori ( 1960 ) . in this area , the main subject of investigation is the urban fabric of the city at the scale of the neighborhood , street , plot and building .
a different branch of urban morphology has stemmed from the sciences of complex systems building on a long standing tradition in regional analysis , economic geography and modeling @xcite .
complexity in the built environment is here investigated with the same instruments used for other classes of self - organized phenomena in nature , technology and society @xcite .
these works are now flanked by a growing interest on complex spatial networks within the community of physics @xcite .
spatial networks and in particular planar graphs are suited to model a number of real phenomena @xcite .
we are here interested in the study of a particular class of spatial networks that describe the street pattern of cities .
the beginning of these studies can be traced back to the classical works on regional transportation networks based on graph theory @xcite .
the advent of complex system science and its paradigm @xcite , jointly with the increasing availability of spatial and time geo - referenced data , has given a new boost to these studies , and several important contributions have appeared recently @xcite .
in @xcite , masucci _ et al . _ study the structural property of the london street network in its dual and primal representation . in @xcite , the autors , by using 40 urban networks in a dual representation , found a small - world structure and a scale - free property for both street length and connectivity degree , and used various centrality indices as indicators of the importance of streets .
et al . _ developed a comparative analysis of the betweenness distribution in 20 cities in germany , suggesting a relation with vehicular traffic @xcite .
others have focused on centralities in primal and dual representations of street networks @xcite , and on other structural features , such as the number of cycles of a given length @xcite .
however , an important and still open question in urban morphology has to do with the characterization of classes of cities based on their form .
this is a preliminary step to approach a comparative analysis aimed at the classification of cities . in this paper
we propose a classification based on the distribution of street centrality by cross - comparing real cases and therefore with no use of null models .
we limit our study to the characterization of city form without exploring its impact on collective behaviors , an area of research which , at the scale of entire cities , is now finding new opportunities through the exploitation of massive datasets from online geo - social networks and mobile geo - referenced systems @xcite in our study , we first observe the geometry of the networks following the approach recently suggested in @xcite ; we consider the distribution of basic indices of the primal street networks such as street angles at intersections and street lengths .
secondly , we investigate four different centralities indices computed over the entire urban street network , highlighting their distribution and their mutual correlation .
we considered 10 european cities namely edinburgh , leicester , sheffield , oxford , worcester , lancaster , catania , barcelona , bologna and geneva .
we show that these cities share some structural geometric properties , which are the same found in other planar spatial systems such as those of leaves @xcite , which suggests that planarity in itself is a driver across spatial systems in various domains .
however , we also show that cities are different .
we witness that some cities , like catania , stand alone in terms of basic geometric properties and , more importantly , that the various distributions of centrality are reconcilable to one common pattern ( a power law ) only if a largely minoritarian subset of streets are taken into consideration .
these results have to do with the extreme heterogeneity of the cities visible traits as resulting from the interplay of entirely different phenomenon in time , such as historical accidents ( including planning ) , physical constraints or just random events @xcite .
in particular , we show how cases tend to cluster in groups after the whole set of centrality measures distributions are considered in a way that suggests major planning interventions and physical geographic constraints are key .
the paper is organized as follows . in section [ s : basicprops ]
the case studies are presented and their geometric properties are introduced and analyzed . in section [
s : centralities ] the study of the four centrality indices is illustrated along with the clusterization of cases according to their combined behavior as a result of the application of a principal component analysis ( pca ) of the distribution of those four centralities .
finally in section [ s : discussion ] we offer a discussion of the results and our conclusions .
we address the analysis of 10 european cities , namely : edinburgh , leicester , sheffield , oxford , worcester and lancaster ( uk ) ; catania and bologna ( it ) ; barcelona ( es ) ; geneva ( ch ) ( fig .
[ f : map ] ) . these cities are variously located and present remarkably different economic , cultural and climatic conditions along with great variety of characteristics like population and size ( tab .
[ t : ppts ] ) .
we represent the street network of these cities such that intersections are nodes and streets are links between nodes , i.e. a primal representation @xcite .
we analyse such networks in terms of their basic properties and several geometric indices such as street length and the angle they form at intersections . before getting deeper into the analysis ,
we introduce some basic concepts of graph theory . a graph ( or network )
is a mathematical object which consists of two sets : @xmath0 and @xmath1 .
the @xmath2 elements of the former are called _ nodes _ , while the @xmath3 elements of the latter ( unordered couple of nodes ) are called _ links_. + there are many ways to represent a graph , but , the most common one is the _ adjacency matrix _ @xmath4 , a @xmath5 square matrix whose entry @xmath6 is equal to one if link between nodes @xmath7 and @xmath8 exists and zero otherwise .
the _ degree _ of a node @xmath7 , @xmath9 , is the number of links incident with it .
the _ average degree _
@xmath10 , is nothing but the average of the degrees over all the nodes in the network .
networks of street patterns belong to a particular class of graphs called , _ planar graph _ , i.e. graphs whose links cross only on nodes . in our case ,
the nodes represent street intersections , while the links are the streets centerline , a network made using this convention is called _ primal network _ ( fig .
[ f : nets ] ) .
our networks are also weighted and each link ( @xmath7,@xmath8 ) carries a numerical value @xmath11 expressing the intensity of the connection . the natural choice , in our case , for the functional form expressing the weight of a link connecting nodes , say @xmath7 and @xmath8 , is to put @xmath11 equal to the length of the street connecting , @xmath12 .
our analysis starts by importing the street system into a geographical information system ( gis ) environment .
data of the street systems have been retrieved from different sources : for example , in all uk cases we have used the ordnance survey maps , while in italian cases we have used data from city councils planning offices , and in barcelona we worked on a dataset provided by the agncia decologia urbana de barcelona .
given that these geographical street networks had been mainly built for the sake of traffic navigation or planning , they presented characteristics that not always fitted the purposes of a centrality analysis ; for example , multi - lane streets were usually represented with one link per lane rather than one link per street . as a result
we have prepared our databases by first cleaning the networks accurately to remove link redundancies , fix short missing links , collapse unconnected links on the same node when needed , continuously confronting the networks with aerial images of the real cities drawn from remote sensing sources such as google earth .
such procedure was undertaken both manually and through ad - hoc tools in a gis environment .
for the definition of the boundary of the urban systems , we followed the border of the built - up area extending it by roughly 1 km outbound . considering the entries of tab .
[ t : ppts ] , we can see how various selected cities are for example in terms of size , from small cases like lancaster to large ones like barcelona , or in terms of street intersection density , from very dense cities like catania to more sprawled ones like edinburgh .
we have selected cities with different levels of geographic constraints , from those like geneva and oxford traversed or limited by large natural water features to those like catania and bologna that sit on uninterrupted plains , and a different prominence of planning history , from those self - organized or only fragmentarily planned like leicester or bologna to one like barcelona whose street layout had been heavily determined by one single planning vision ( the 1859 cerd plan ) .
we see that the particular planning history of barcelona is reflected in the low values of both the standard deviation of the street length and the percentage of dead - end streets on the total number of streets , both resulting from the extensive adoption of a rigid homogeneous grid layout .
the extreme diversity of selected cases has been pursued in order to make the comparative analysis of similarities and differences more profound . [ cols="<,^,^,^,^,^,^,^,^,^,^",options="header " , ] streets are not always straight lines . in order to study the distribution of angles formed by streets at intersections
we must use an equivalent network in which all the streets are represented by straight lines ( i.e. substituting the curved streets with straight ones ) and where the link weight is equal to the euclidean distance between its end nodes .
we name this network _
euclidean network_. as for all the generalization models , results should be interpreted with caution because of the effect that the approximation may produce on the real structure of the network . such caution suggests a preliminary test .
the inset in fig .
[ f : angles ] shows small divergences between the distribution of street length in the original and the euclidean networks for the city of leicester . such simple
test confirms that the euclidean generalization does not lose relevant information , i.e. that streets in cities are not always straight , but predominantly so , confirming the finding provided in @xcite .
therefore we are confident to study the distribution of the angles formed by street intersections in the euclidean network as a reliable approximation of the original .
the distributions of the angles formed by streets at intersections are shown in fig .
[ f : angles ] . at a glance
, we note that all the cities share the same behavior exhibiting a double peak shaped distribution around the characteristic values of 90 and 180 degrees , respectively .
this finding confirms the analogous results that have been found in other kinds of spatial planar networks , like those of leaf venations @xcite , as a result of tensorial stress fields or simple force models . the incidence of certain motifs in complex networks is a long standing object of investigation , @xcite .
here we want to focus not on the cycle s shape and quantity , but on the relation between the total length of dead links within a given cycle and its area .
even if we are measuring static systems , i.e. systems that do not change in time , we should remember that a cycle is the result of an evolutionary process that starts with short dead end streets sprouting from the longest edges of the cycles and then extending towards the opposite edge until splitting the original cycle in two smaller sub - cycles .
dead end streets can be interpreted as sprouts of new cycles in parcels still subjected to evolution or as crystallized fractures that do not undergo further development .
their quantity is given by the index @xmath13 shown in tab .
[ t : ppts ] and it can be thought of as an estimator of the abundance of cycles in the intermediate evolutionary stage of their lifetime as suggested in @xcite .
such assumptions are supported by the result shown in fig .
[ f : deadends ] , where we report the sum of the length of dead links inside each cycle versus the area of the cycle itself for each city .
the distribution shows a clear power - law behavior with a common exponent close to 0.8 .
it is worth to note that the power - law behavior is not affected by any factor like the fraction of dead ends @xmath13 , or the average degree @xmath14 .
similar results have appeared in @xcite .
of course , our findings can be truly confirmed only by investigating the evolution of urban streets in time with the support of empirical data .
the concept of centrality has been used for many years in network and social science and , starting from the seminal work by wasserman @xcite , there has been a growth of literature both regarding centrality on social networks as well as other kinds of networks .
depending on the definition , centrality can be understood as meaning proximity between nodes , accessibility from other nodes , or being in a strategic position for connecting couple of nodes .
it is clear that from different definitions of centrality , a node actor can be placed at different centrality ranks and that the same node can result with high value for a centrality while yielding weak values for another one .
therefore , for different cities , we can reasonably expect slightly different distributions of centralities .
moreover , we can identify how centralities are mutually dependent and correlated .
what makes centralities particularly suited for geographical studies is that they can be visualized and mapped .
we are interested in understanding if and how these ranges of correlations and fluctuation can help to classify cities that share the main network morphology .
for example , a grid - like network can be different from a radial one , but only looking at the statistical distribution of centralities might not be enough for a proper classification .
this problem has been already analysed by crucitti and porta @xcite where the statistical distribution of closeness , betweenness , straightness and information centralities have been analysed for a sample of twenty city parcels of one square mile .
_ found significant differences between cities and through a cluster analysis they proposed a classification of different urban patterns .
the approach proved to be effective in capturing essential features of urban form as emerging in limited samples selected for their inner morphological consistency , but dealing with entire cities poses the problem of the classification of internally complex objects predominantly composed of different parts each possibly exhibiting different properties .
so , the question about the validity of such a procedure on a whole city still need a response . in order to validate
if centrality indices can be used in the classification of entire cities , we propose here a clustering method based on the principal component analysis ( pca ) made on the distribution of centralities and on their moments . before discussing this part of the research
, we must introduce briefly the adopted centrality indices .
* betweenness centrality * , @xmath15 , is based on the idea that a node is more central when it is traversed by a larger number of the shortest paths connecting all couple of nodes in the network . more precisely , the betweenness of a node @xmath7 is defined as in @xcite : @xmath16 where @xmath17 is the number of shortest paths connecting @xmath8 and @xmath18 , while @xmath19 is the number of shortest paths connecting @xmath8 and @xmath18 and passing through @xmath7 .
* straightness centrality * , @xmath20 , originates from the idea that the efficiency in the communication between two nodes @xmath7 and @xmath8 is equal to the inverse of the shortest path length , or geodesic , @xmath21 @xcite . in the case of a spatial network embedded into a euclidean space
, the straightness centrality of node @xmath7 is defined as : @xmath22 where @xmath23 is the euclidean distance between nodes @xmath7 and @xmath8 along a straight line .
this measurement captures to which extent the connecting route between nodes @xmath7 and @xmath8 , let s say between each street junctions , deviates from a virtual straight route .
the * closeness centrality * , @xmath24,of a node @xmath7 is based on the concept of minimum distance , in topological sense , i.e. the minimum number of edges traversed to get from @xmath7 to @xmath8 + @xcite and is defined as in @xcite : @xmath25 where @xmath26 is the average distance from @xmath7 to all the other nodes .
closeness centrality is a classical centrality index that has been widely used in urban geography and econometrics as well as in regional planning , where it gives an idea of the cost that spatial distance loads on many different kinds of relationships that take place between places , people , activities and markets .
the * accessibility * @xmath27 is a measure recently introduced by travenolo and costa + @xcite .
it has been used for studying the property of very different spatial networks . in the case of urban networks
, it has been used to investigate the relationship between subway and road systems @xcite .
in addition , the accessibility has been found to be closely related to the borders of networks @xcite , in the sense that nodes with low accessibility tend to belong to these borders .
the @xmath27 of node @xmath7 measures the ratio of neighbouring nodes that are effectively reached by an agent randomly navigating the network against the actual number of nodes that belong to the neighborhood .
more precisely , @xmath27 takes into account the number of nodes effectively accessed by each node of the network , as well as the probabilities of such accesses .
first , we evaluate the transition probability @xmath28 which describes the probability for an agent leaving from node @xmath7 to reach node @xmath8 after @xmath29 steps along a given type of walk . at each step
, the agent located at node @xmath30 , chooses a random neighbor of @xmath30 and jump to it .
these rules define a random walk over the network .
when the transition probabilities are very heterogeneous , we have low values of accessibility , meaning that the random walks are biased toward a certain number of nodes , which is , less than the number of nodes which can be reached after @xmath29 steps . on the other hand , when the transition probabilities are homogeneous , all nodes which can be reached after @xmath29 steps are accessed , on average , the same number of times .
this case corresponds to the highest values of accessibility .
the heterogeneity of the transition probabilities is quantified in terms of the classical concept of entropy , so that the mathematical definition of accessibility , of a node @xmath7 , with respect to the number of steps @xmath29 is given as @xmath31,\ ] ] also , we have considered the transition probability for unitary step @xmath32 , as : @xmath33 where @xmath34 is the weight of the edge @xmath35 . in order to take the geography into account
, we considered @xmath36 .
for disconnected nodes , we assume @xmath37 such that @xmath38 . in order to investigate
if the distribution of centralities can describe the main geographical differences within cities , we use the pca approach .
this well - established method of multivariate statistics implements a projection of the distribution of objects ( in our case , cities ) from a higher into a lower dimensional space such that the maximum dispersion of the data is observed at the first new axis ( or principal variable ) , and so forth .
this projection is optimal in the sense of fully decorrelating the original data , therefore removing all correlations between the original measurements describing the objects .
so , since the data dispersion is better described by the first principal axes , the remainder axes can be discarded .
pca is therefore particularly relevant to the presents studies because : ( i ) it decorrelates the original measurements ; ( ii ) it provides a projection of the data that maximizes their dispersion ( i.e. the differences between the cities ) ; and ( iii ) it allows the visualization of the distribution of objects ( when projected into 2d or 3d spaces ) .
the pca consists in obtaining the covariance matrix of the original data and then extracting its eigenvalues and respective eigenvectors .
the eigenvalues can be shown to correspond to the variances along each new axis , and the each respective eigenvector component provides the coefficient of the linear combination of the original measurements used to project the original data into the respective axis .
therefore , the effectiveness of pca in projecting the data can be inferred by inspecting the eigenvalues in descending order .
for instance , if the two largest eigenvalues account for 75% of the overall variance , it can be understood that these two axes are describing the original distribution of points in an effective way , and that the other axes can be overlooked without losing much information .
first , we evaluate the histogram of each centrality measurement considering 20 bins . in this way
, each dimension corresponds to the relative frequency of centrality values in a small range .
the histogram of each one of the four centralities were merger together in order to create an eighty dimension feature vector for each city .
for instance , figure [ f : pcacontrib ] ( top ) shows the feature vector obtained for the city of leicester .
, width=302 ] in the second approach , the feature vectors were created by considering the 20th first moments of the centralities distributions merged together , also generating an eighty dimension vector .
it is important to note that , both first and second approaches provided similar results .
finally , the dimension of the feature vectors were reduced from eighty to two , in which each original dimension has contribution according the values shown in figure [ f : pcacontrib ] ( bottom ) and fig .
[ f : pcacontrib ] ( top ) shows the explanation of each dimension . by plotting the first two dimensions , it is possible to account for almost 50% of the dispersion , while a clear differentiation between cities can be seen in fig .
[ f : pca ] .
since it is impossible to visualize the high - dimensional data so as to try to recognize the clusters , we used an agglomerative clustering method , the so - called complete linkage method to perform this task . in this method ,
the euclidean distance between two clusters is given by the value of the shortest distance between any object belonging to these clusters .
the final result is shown in fig .
[ f : dend ] as a dendrogram , where the colours correspond to the two clusters identified by using a threshold parameter 0.7 of the maximum distance between two cities .
this figure clearly shows the emergence of one cluster of cities ( grey color ) and of a group of cities very different each others but still belonging to a single family ( black colour ) , so we may intepret this agglomeration as the existence of two main different classes or typologies of cities . in this classification a clear group composed by catania , bologna , sheffield , edinburgh and leicester is separated from geneve , lancaster , oxford , worcester and barcelona .
though the reasons behind this separation may be manifold and it needs further experimental tests , we notice that comparing the plans of those cities and the topography of their terrain it is possible to conclude that physical constraints can be a key factor to classify cities .
we have found that cities traversed by rivers or bordered by a lake were separated from those that the growth has never been boundend or divided by physical constraints .
the exeption of barcelona , a city with a major planning events ( the plan cerd ) confirms this hypothesis posing the great urban planning operations in the list of the geographical constraints .
in this research we analyse the street network of 10 european cities represented in a primal way , where intersections are translated into the nodes of a network and the connecting streets into the links .
we study first the geometric properties of the networks and then the way centrality is distributed over them according to four different definition of centrality .
we show that selected cities share several universal geometric patterns , such as the average street length ( which is limited within a remarkably narrow range of values ) , the distribution of angles between streets at intersections , and the distribution of the total length of dead end streets as a function of the area of the cycle they belong to .
in addition , we confirm that the distribution of street lengths follows universal power law behavior in the long tail of the dataset , i.e. for high values of street length . however , we highlight that the conventional way to represent the distribution of street lengths in search of the power law behavior leaves out of the picture the vast majority of the data , i.e. the number of streets whose length distribution can not be accurately fitted by a power law function is in the order of the 90% of the whole dataset .
we therefore investigated the actual behaviour of the street length distribution in a semi - log scale finding the emergence of remarkably different behaviours that seemingly reflect the diversity of local history and conditions .
finally , we found that the distribution of the four centrality indices over the street networks allows us to clearly characterize two different clusters of cities that appear to be predominantly informed by major topographic and geographic local features like the presence of rivers or lakes .
we also found that planning events of extreme magnitude like the occurrence of the plan cerd in barcelona may account for the uniqueness of this city as measured through the distribution of street centrality .
we show how the analysis of simple geometric properties of the street network as well as a more complex evaluation of the street centrality can highlight differences between cities and how these differences can be used for classifying cities in categories .
it is clear that global geometric characteristics of the street network exhibit `` universal '' behaviors but it is reasonable to argue that , as these behaviors equally emerge in a vast range of transportation networks in nature and technology that like the street networks are mainly planar , the universality of those features may be related to the planarity of the systems rather than a particular `` nature '' of cities as such .
regarding the analysis of the morphology of cities , we believe our study highlights that if the universal rules governing the evolution of planar hierarchical networks are not discriminated , the extreme differences of their inner urban structure may be easily underestimated and , from a qualitative point of view , it may leave local patterns out of the picture .
we focused on this discrimination process showing how to operate at global level in order to highlight local patterns where the extreme diversity of our cities emerges and should be accounted for , especially when we move on to the problem of classification , i.e. that of finding a taxonomy for urban types .
the authors thanks shibu raman , tim jones , mike jenks and colin pooley for providing the data of some of the analyzed cities .
emanuele strano is grateful to marc barthlemy , salvatore scellato and andrea perna for their suggestions and help and to the faculty of engineer at university of strathclyde for his mres grant .
luciano is grateful to fapesp ( 05/00587 - 5 ) and cnpq ( 301303/06 - 1 and 573583/2008 - 0 ) for financial support .
matheus viana thanks fapesp ( 2010/16310 - 0 ) for his phd grant .
|
we compare the structural properties of the street networks of ten different european cities using their primal representation . we investigate the properties of the geometry of the networks and a set of centrality measures highlighting differences and similarities among cases .
in particular , we found that cities share structural similarities due to their quasi planarity but that there are also several distinctive geometrical proprieties .
a principal component analysis is also performed on the distributions of centralities and their respective moments , which is used to find distinctive characteristics by which we can classify cities into families .
we believe that , beyond the improvement of the empirical knowledge on streets network proprieties , our findings can open new perspectives in the scientific relation between city planning and complex networks , stimulating the debate on the effectiveness of the set of knowledge that statistical physics can contribute for city planning and urban morphology studies . + * keywords : * complex street networks , urban form , city classification , centrality .
|
to obtain the reaction rates relevant for the nuclear astrophysics , experimental data should be extrapolated to the very low - energy region ( the gamow window ) .
the cross section depends strongly on the energy and therefore is expressed in terms of the astrophysical @xmath13 factor @xmath14 in the system of units , in which @xmath15 , the coulomb ( sommerfeld ) parameter @xmath16 .
@xmath17 is the charge of the nucleus @xmath18 , @xmath19 and @xmath20 are the relative momentum and the reduced mass of the interacting nuclei , @xmath21 is their relative kinetic energy in the c.m . frame .
the extrapolation of the cross section down to low energies assumes , that the coulomb potential of the target nucleus and a projectile results from bare nuclei . in experimental conditions , however , the coulomb potential is screened by electrons surrounding the target nucleus , thus reducing the height of the coulomb barrier and leading to a higher cross section .
as the energy approaches zero , the electron screening potential @xmath10 enhances the bare nucleus astrophysical factor @xmath22 .
the estimated electron screening potential for the @xmath0li@xmath2he reaction in the adiabatic limit is a difference in atomic binding energies between li and be@xmath23 , that is 186 ev @xcite . the trojan horse ( th ) experiment
has indirectly measured the bare nucleus astrophysical factor @xcite .
a comparison of the th results and a direct measurement led to @xmath24 ev @xcite .
large @xmath10 values were obtained by engstler et al .
@xcite using the polynomial fits , as well as by barker @xcite using the r - matrix fits .
barker also noted @xcite , that fixing @xmath10 at 175 ev results in a reasonable fit with only slightly higher @xmath25 .
ruprecht et al .
@xcite reported @xmath26 ev and concluded , that the lower screening energy is due to the influence of the @xmath5 subthreshold resonance .
the current work presents a new r - matrix analysis for the low - energy @xmath0li@xmath2he reaction , that considers three subthreshold resonances . the largest contribution to
the low - energy @xmath13 factor comes from the resonances and the subthreshold resonances closest to the threshold of the compound nucleus .
the three subthreshold resonances closest to the threshold of @xmath27be are the @xmath28 subthreshold state at @xmath29 kev , followed by the @xmath30 subthreshold resonance at @xmath31 mev and a @xmath28 state at @xmath32 mev . the @xmath28 resonance at @xmath33 mev is not included in the calculation .
the astrophysical @xmath3 factor is first calculated including only the @xmath5 subthreshold resonance at @xmath29 kev , and then by adding two more subthreshold resonances .
the fitting to the direct low - energy experimental data @xcite is done using the nonlinear least - squares procedure .
golovkov et al .
@xcite data are the outliers , and , hence , are not included in the calculation .
the errors of the experimental data of engstler et al .
@xcite do not include the reported uncertainties arising from the number of counts , angular distributions , the effective energy and the target stoichiometry , and , hence , are reduced by @xmath34 .
including the latter errors would result in a normalized @xmath35 . here
, the normalized @xmath36 is defined as @xmath37 , where @xmath38 is the number of experimental points used in the fit calculation , and @xmath39 is the number of parameters .
the error bars of elwyn et al .
experimental data @xcite are enlarged to get @xmath40 , as the underestimated errors may lead to a bias in the derived slope .
the chosen error bars for elwyn et al .
data @xcite are set to the @xmath41 of the measured @xmath3-factor value .
the r - matrix fits are calculated using the modified r - matrix formulas @xcite , which use the `` observed '' rather than the formal parameters .
the alternate parametrization allows to set the resonances energies at the experimental values , instead of conventionally @xcite choosing the random formal energies in the vicinity of the resonance , and calculating the resulting `` observed '' energies after the r - matrix fit . in the three - level fit ,
the two @xmath42 resonances interfere , and the two - level r - matrix equations are used .
the contribution of the @xmath43 state is added incoherently .
considering only two channels for each state , the low - energy astrophysical @xmath13 factor is defined as @xcite @xmath44 here , the statistical spin factor @xmath45 is @xmath46 the collision matrix @xmath47 in terms of the `` observed '' parameters is identical to the one , that is expressed in terms of the formal parameters @xcite @xmath48 and the @xmath49 matrix in terms of the `` observed '' parameters is defined as @xmath50 here , @xmath51 is the shift function @xmath52 , evaluated at energy @xmath53 , @xmath54 is the penetribility , and @xmath55 .
the r - matrix formula of lane and thomas @xcite requires an inclusion of all partial waves . for the initial @xmath5 state one would have @xmath56 , @xmath57 , @xmath58 , @xmath59 partial waves and @xmath60 , @xmath59 partial waves for the initial @xmath61 state .
the exit channel for the @xmath42 state is @xmath62 , and , for the @xmath61 state , it is @xmath63 .
the asymptotic normalization coefficients ( anc s ) for the bound states , as well as the reduced width amplitudes @xmath64 s for the final channels are unknown and , therefore , are treated as parameters in the r - matrix calculation . assuming that the @xmath65-wave approximation for the incoming deuteron is reasonable at very low deuteron energies , the number of free parameters reduces significantly .
unconstrained fits to direct data do not allow reliable determination of all parameters ; therefore present calculations use constrained fits , which are subject to assumptions of the physical parameters .
first , the r - matrix fits depend on the restrictions placed on the experimentally unknown @xmath66 channels partial widths .
the subthreshold resonances under consideration are broad .
the experimental resonance widths of @xmath67 are 0.8 mev and 0.88 mev for the @xmath42 states at 22.2 mev and 20.1 mev , respectively , and 0.72 mev for the @xmath61 state at 20.2 mev @xcite .
the total width for each level is a sum of all partial widths .
the threshold of the @xmath68li@xmath1he reaction entrance channel corresponds to a high excitation energy in a compound nucleus , hence , many reaction channels are open . considering only the @xmath68li@xmath2he reaction ,
the major contribution to the total width at the subthreshold energy comes from the @xmath69 channel due to the large q value of the reaction @xmath70q@xmath71 mev@xmath72 and varies slowly with energy .
the relative widths @xmath73 of the @xmath42 subthreshold resonance at @xmath8 mev have been determined experimentally to be @xmath74 @xcite .
@xmath75 has been reported @xcite for the @xmath4 state .
page @xcite performed a many - level multichannel simultanous r - matrix fit to known @xmath66 elastic scattering data , as well as @xmath76 and other channels data .
a single - level fit to data , as well as the fit , that includes all three aforementioned subthreshold resonances , fails for the bare @xmath0li@xmath2he astrophysical @xmath3 factor using the suggested formal @xmath66 channels partial widths @xmath77 mev for the @xmath42 subthreshold state at 22.2 mev ( experimental total width 0.8 mev ) , @xmath78 mev for the @xmath61 subthreshold state at 20.2 mev ( experimental total width 0.72 mev ) , and @xmath79 mev for the @xmath42 subthreshold state at 20.1 mev ( experimental total width 0.88 mev ) in ref . @xcite .
the fit also fails , if one includes an arbitrary background .
reference @xcite notes that because the resonances are broad , the resonance contribution can not be separated from the background contribution and therefore the elastic scattering data do not provide accurate resonance parameters . also , the best fit of ref .
@xcite includes an additional unknown level at 580 kev ( excitation energy 22.78 mev ) with the formal @xmath80 mev and @xmath81 mev for the @xmath82-wave deuteron .
the relative width @xmath83 for the @xmath5 subthreshold resonance at @xmath8 mev , found in @xcite , is significantly smaller than that given in literature @xcite and obtained from the @xmath84li@xmath85 and @xmath86be@xmath85 reactions data . also , the @xmath87 of the @xmath5 subthreshold resonance at @xmath6 mev obtained by ref .
@xcite is much smaller than that obtained by other r - matrix or polynomial fits to data @xcite .
hence , the present study considers only those constrains , that are imposed by the experimental measurements . for
the bound states the anc s are related to the reduced width amplitudes @xcite @xmath88 where @xmath89 is a whittaker function . without the experimentally imposed constrains ,
it is not possible to say which range of the anc s is more appropriate .
therefore the anc s are treated as free parameters .
the channel radii associated with the range of the nuclear force are calculated by a conventional formula of lane and thomas @xcite @xmath90 , where @xmath91 and @xmath92 are the mass numbers and @xmath93 is a numerical value between @xmath94 and @xmath95 fm . in principle , the collision matrix is independent of the choice of the channel radii , provided a large enough number of levels is included into the analysis , and , consequentially , is the astrophysical @xmath3 factor .
partial widths and energies of the resonances resulting from an r - matrix fit calculation should also be channel radii independent @xcite .
the sensitivity of the resonance parameters to the adopted channel radii in the initial and final channels is illustrated in table [ tab1 ] for a single - level calculation that uses the @xmath65-wave approximation for the deuteron .
the parameters depend strongly on the @xmath0li+d channel radius , which may indicate a need for the inclusion of the additional initial channel partial waves in the r - matrix fit .
also , as the analysis deals with broad resonances , the strong dependance on channel radii may support a need for an inclusion of additional levels into the r - matrix fit .
ccccccc @xmath96 & @xmath97 & @xmath98 & @xmath99(mev ) & @xmath100 & @xmath36 & @xmath10(ev ) + + 4.5 & 4.0 & 22.2610 & 0.6952 & 22.5222 & 0.7678 & 76.9987 + 4.5 & 4.5 & 22.2608 & 0.6947 & 22.5279 & 0.7681 & 76.5234 + 4.5 & 5.0 & 22.2606 & 0.6943 & 22.5278 & 0.7682 & 76.6777 + 4.5 & 5.5 & 22.2605 & 0.6940 & 22.5309 & 0.7684 & 76.4070 + 5.0 & 4.0 & 22.1911 & 0.7928 & 22.4021 & 0.7494 & 83.2557 + 5.0 & 4.5 & 22.1907 & 0.7923 & 22.4066 & 0.7496 & 82.8810 + 5.0 & 5.0 & 22.1904 & 0.7919 & 22.4087 & 0.7497 & 82.7473 + 5.0 & 5.5 & 22.1902 & 0.7917 & 22.4115 & 0.7498 & 82.4966 + 5.5 & 4.0 & 22.1368 & 0.9330 & 22.2113 & 0.7303 & 92.8859 + 5.5 & 4.5 & 22.1362 & 0.9327 & 22.2158 & 0.7305 & 92.4693 + 5.5 & 5.0 & 22.1357 & 0.9325 & 22.2155 & 0.7306 & 92.6005 + 5.5 & 5.5 & 22.1354 & 0.9323 & 22.2184 & 0.7307 & 92.3145 + when allowing the channel radius to vary as one of the parameters , the single - level best fit sets the initial channel radius @xmath101 fm .
the best fit places the subthreshold resonance at 22.252 mev with @xmath102 mev .
the corresponding @xmath103 ev . there is no obvious reason , however , to set the initial channel radius to this value , as one looks for a range of radii in which the conclusions of the calculation are reasonably stable , rather than for a single value which produces the lowest @xmath25 .
in this analysis , the chosen values for the initial and final channel radii are 5.0 and 4.5 fm , respectively .
the sensitivity to the initial channel radius is evaluated in the error bars . as deviations of the conclusions
could be attributed to the effects of other levels , the single - level @xmath104-matrix fit for the @xmath5 subthreshold resonance and the @xmath104-matrix fit , that includes the three aforementioned subthreshold states , are considered .
the calculations use the @xmath65-wave approximation for the deuteron , as with the existing data including more partial waves would only introduce more unknown fitting parameters .
an unrestricted single - level @xmath104-matrix best fit to the low - energy experimental data @xcite that uses the energy @xmath105 of a @xmath5 subthreshold resonance near the 22.2 mev excitation energy , the electron screening potential @xmath10 , as well as the reduced width amplitudes @xmath64 and @xmath106 as free parameters , is shown in the top panel of fig .
[ figs1 ] .
the best fit places the subthreshold resonance at @xmath107 mev with @xmath108 mev .
the resulting bare astrophysical @xmath3 factor at zero energy @xmath109 mev b and the electron screening @xmath110 ev .
the @xmath25 of the fit , that uses four parameters and 82 data points , is @xmath111 .
the normalized @xmath36 is then @xmath112 .
the subthreshold energy of the @xmath5 state is then fixed at the experimental value of @xmath29 kev ( excitation energy 22.2 mev ) and the partial width of the @xmath69 channel is restricted not to exceed the total width of the state .
the best fit that treats @xmath10 , @xmath64 and @xmath106 as parameters , provides @xmath113 mev b , @xmath114 mev and the electron screening @xmath115 ev . the @xmath116 is only slightly higher than that resulting from the former fit , in which the subthreshold resonance energy is allowed to vary , resulting in a slightly smaller @xmath117 .
the ambiguity of the fit due to the sensitivity to the choice of the channel radii is larger for the restricted fit .
the larger errors correspond to the fits for which the initial channel radius was reduced , resulting in larger values of the electron screening , the larger @xmath118 partial width and the smaller @xmath119 .
to neglect the electron screening , the single - level r matrix is fit to data above @xmath120 kev @xcite .
the calculation for the bare @xmath3 factor is done treating @xmath64 and @xmath106 as the only parameters .
the observed @xmath3 factor is then obtained by varying @xmath10 as a single parameter to fit data below 1 mev @xcite . the best fit is shown in the bottom panel of fig . [ figs1 ] .
resulting @xmath121 mev b and @xmath122 mev .
@xmath25 of the best fit to 50 data points is 66.6727 , @xmath123 .
the reduced width amplitudes are @xmath124 mev@xmath125 and @xmath126 mev@xmath125 .
the corresponding anc is @xmath127 fm@xmath125 .
resultant @xmath129 ev .
the sensitivity for the latter fit tracks the same way as the previous fit : reducing the channel radius strongly increases the electron screening .
also , the single - level @xmath104-matrix best fit at a fixed energy of the @xmath5 subthreshold resonance at -80 kev , results in different sets of parameters , when fitting the electron screening simultaneously with other parameters , and , when the electron screening is neglected below 60 kev .
fitting all data below 1 mev with the screening potential @xmath10 included in the set of the fitting parameters , results in @xmath130 ev , while fitting data at energies above 60 kev , where supposedly the electron screening can be neglected , and then using the @xmath10 as a single fitting parameter to fit all data below 1 mev , results in the nearly twice as low screening potential @xmath131 ev .
the second closest subthreshold resonance to the threshold of @xmath27be ( excitation energy 22.28 mev ) is the @xmath30 state at -2.08 mev ( excitation energy 20.2 mev ) .
the @xmath3 factors for the two states are added incoherently .
the @xmath104-matrix best fit , that treats the @xmath64 s and the anc s as free parameters , results in a much too large width of the @xmath30 state . restricting the partial widths of the @xmath66 channels not to exceed the total widths of each state leads to @xmath132 mev b , @xmath133 mev , @xmath134 mev .
the best fit parameters are anc @xmath135 fm@xmath125 and @xmath136 mev@xmath125 for the @xmath5 , and anc @xmath137 fm@xmath125 and @xmath138 mev@xmath125 for the @xmath4 state .
@xmath25 of the fit , that uses four parameters and 50 data points , is 65.8849 , @xmath139 .
the electron screening is then fit to data below 1 mev as a single parameter , resulting in @xmath140 ev . by restricting the @xmath99 of the @xmath30 subthreshold resonance not to exceed the half of the total width , as reported in @xcite ,
the best fit parameters become anc @xmath141 fm@xmath125 and @xmath142 mev@xmath125 for the @xmath5 , and anc @xmath143 fm@xmath125 and @xmath144 mev@xmath125 for the @xmath4 state .
@xmath25 of the best fit is 66.6689 and @xmath145 .
then @xmath146 mev b , @xmath147 mev , @xmath148 mev , and the electron screening @xmath10 is @xmath149 ev .
thus , adding the second subthreshold resonance hardly affects the extracted screening potential from that obtained for a single subthreshold state .
it also does not reduce the sensitivity to the channel radii .
one notes that whether a single @xmath5 subthreshold state is used in the @xmath104-matrix fit , or the two subthreshold states @xmath5 and @xmath4 are considered , the determined alpha partial width of the @xmath5 state is significantly larger than the one obtained in ref . @xcite . in this section ,
the @xmath104-matrix fit considers three subthreshold resonances : two @xmath42 and one @xmath43 .
the @xmath42 states are interfering and the @xmath43 state is added incoherently .
the partial widths of the @xmath66 channels for each state are constrained not to exceed the total experimental widths . in the absence of
further constrains , the @xmath104-matrix best fit to the low - energy experimental data @xcite for the bare astrophysical factor between 60 kev and @xmath150 mev results in @xmath151 mev b. the best fit parameters are shown in table [ tabpar ] as a set 1 .
the partial widths of the @xmath66 channels are @xmath152 mev , @xmath153 mev and @xmath154 mev for the subthreshold levels at 22.2 mev , 20.2 mev and 20.1 mev respectively .
@xmath25 of the fit , that uses six parameters and 50 data points , is 65.5116 , @xmath155 .
the electron screening is then fit to all data below 1 mev as a single parameter , resulting in @xmath156 ev .
the best fit looks identical to the panel ( b ) of fig .
[ figs1 ] .
cccccccccc & & & & & & + @xmath157 & @xmath158 & & & & & & & & + & & c ( fm@xmath125 ) & @xmath159 & @xmath10 ( ev)&@xmath160 & c ( fm@xmath125 ) & @xmath159 & @xmath10 ( ev)&@xmath160 + + @xmath161 & 22.2 & 2.2607 & 0.2420 & & 0.6932 & 2.2378 & 0.2558 & & 0.7746 + @xmath162 & 20.2 & 1.7430 & 0.2009 & & 0.4892 & 1.2790 & 0.0366 & & 0.0163 + @xmath161 & 20.1 & 2.4390 & 0.1146&71.393 & 0.1462 & 2.1159 & 0.0925 & 149.521 & 0.0952 + & & & & & & + the sensitivity of the fit to the choice of channel radii for the three level fit tracks the same way as the sensitivity for a single - level fit , possibly indicating that this problem is not due to the lack of the background levels .
@xmath163 for the three - level fit is only slightly smaller than the @xmath25 of a single level fit , producing a slightly higher @xmath36 .
due to a strong sensitivity to the channel radii , it is difficult to precisely determine the partial widths of the subthreshold resonances ; however , it is clear that the fit prefers a partial width of the @xmath5 state at 20.1 mev significantly lower than the experimental total width 0.88 mev @xcite and its effect on the @xmath3 factor is small .
the @xmath104-matrix best fit to the low - energy experimental data @xcite that fits the anc s , the reduced width amplitudes @xmath64 s and the electron screening simultaneously , looks very similar to the panel ( a ) of fig .
[ figs1 ] .
s@xmath164 of the best fit is @xmath165 mev b and @xmath166 ev .
the best fit parameters are shown in table [ tabpar ] as a set 2 .
resulting partial widths of @xmath66 channels are @xmath167 mev , @xmath168 mev and @xmath169 mev for the subthreshold levels at 22.2 mev , 20.2 mev and 20.1 mev respectively .
@xmath25 of the fit , that uses 7 parameters and 82 data points is 117.5280 , @xmath170 .
the electron screening potential @xmath171 ev , resulting from the best fit that varies all parameters simultaneously is halved , when the r matrix is fit to the experimental data above 60 kev to neglect the electron screening .
a comparison of the best fit parameters , when @xmath10 is fit simultaneously and separately , listed in table [ tabpar ] , possibly suggests , that the electron screening is not negligible above 60 kev , contrary to ref .
@xcite stating that for the deuteron energies larger than 50 kev the enhancement of the cross section due to the screening effect can be neglected .
therefore , the parameters we recommend are those resulting from a fit that varies @xmath10 simultaneously with other parameters .
the bare @xmath3 factor resulting from different @xmath104-matrix fits is larger than many previously reported calculations @xcite and agrees with s@xmath172 mev b obtained by ruprecht et al . @xcite . while the value of the electron screening potential @xmath10 is sensitive to the choice of channel radii , it is smaller than the adiabatic limit .
we did not observe a significantly larger @xmath10 , as it was reported in ref .
the astrophysical @xmath3 factor for the low - energy @xmath0li@xmath1he reaction dominated by broad subthreshold resonances has been analyzed using the single- , two- and three level @xmath104-matrix fits .
for the low - energy r matrix we use the s - wave approximation for the deuteron . the resulting ambiguity due to
the choice of channel radii is large in a single - level fit , table [ tab1 ] , as well as in the fit that considers three subthreshold resonances .
our goal is to check possibility of determination of the electron screening potential from the low - energy astrophysical @xmath3 factors and to determine the anc s of the subthreshold states and @xmath173 partial widths .
we find that parameters depend on the number of the subthreshold states involved in the fitting .
we consider the fit with three subthreshold states as the most reliable .
we find that the extracted screening potential depends on the used procedure .
if we first fit the @xmath3 factor varying all the parameters at energies above 60 kev , at which according to ref .
@xcite the electron screening potential can be neglected , and then fixing all the parameters and varying only @xmath10 to fit the astrophysical factor at energies below 1 mev , we obtain @xmath174 ev .
however , if we fit the @xmath3 factor at energies below 1 mev varying all the parameters simultaneously including @xmath10 , we get @xmath175 ev .
thus , the result strongly depends on the fitting procedure .
we may conclude that the assumption of ref .
@xcite that electron screening effects are negligible at energies above 50 kev is not valid and our recommended value is @xmath175 ev .
note , that other obtained fitting parameters are also sensitive to the fitting procedure .
our recommended values are the parameters from the set 2 in table [ tabpar ] .
our obtained @xmath173 width for the @xmath42 subthreshold resonance at 22.2 mev @xmath176 mev is higher than the 0.56 mev value obtained by ref .
@xcite and is closer to the 0.76 mev value obtained by @xcite .
the partial width is lower than the total experimental value of 0.8 mev @xcite , which was confirmed by corresponding theoretical calculations .
the @xmath173 partial width for the @xmath61 subthreshold state at 20.2 mev is small and agrees with @xmath75 reported in ref .
the partial width of @xmath42 state at 20.1 mev is much lower than the experimental value of the total width @xcite .
our attempt to fit the @xmath3 factor for the @xmath0li@xmath1he reaction using @xmath177 widths from ref .
@xcite failed and we believe , that the main reason for that is that our @xmath173 partial width for the dominant @xmath42 state at 22.2 mev is significantly higher than 0.11 mev obtained in @xcite .
we conclude that while it is difficult to pinpoint accurately the electron screening potential by fitting existing direct measurements , the obtained electron screening potential is definitely smaller than the adiabatic limit @xmath178 ev @xcite .
this work has been partially supported by the italian ministry of the university under grant rbfr082838 ( firb2008 ) .
a. m. m. acknowledges the support by the us department of energy under grants no .
de - fg02 - 93er40773 , no .
de - fg52 - 09na29467 and nsf under grant no . phy-1415656 .
h. j. assenbaum , k. langake and c. rolfs , z. phys .
a * 327 * , 461 ( 1987 ) .
t. d. shoppa et al .
c * 48 * , 837 ( 1993 ) . c. spitaleri et al . ,
c * 63 * , 055801 ( 2001 ) .
s. engstler et al . , z. phys .
a * 342 * , 471 ( 1992 ) .
f. c. barker , nucl .
a * 707 * , 277 ( 2002 ) .
g. ruprecht et al . , phys . rev .
c * 707 * , 025803 ( 2004 ) .
a. j. elwyn et al .
c * 16 * , 1744 ( 1977 ) .
cai dunjiu , zhou enchen , jiang chenglie 1985 ( unpublished ) .
r. risler et al . , nucl .
phys . a * 286 * , 115 ( 1977 ) .
j. m. f. jeronimo et al . , nucl . phys . * 38 * , 11 ( 1962 ) .
m. s. golovkov et al . , sov .
. phys . * 34 * , 480 ( 1981 ) . c. r. brune , phys
c * 66 * , 044611 ( 2002 ) .
a. m. lane and r. g. thomas , rev .
* 30 * , 257 ( 1958 ) .
d. r. tilley , nucl .
phys . a * 745 * , 155 ( 2004 ) . v. m. pugach et al . , bull .
* 56 * , 1768 ( 1992 ) .
j. benn et al . , nucl .
a * 106 * , 296 ( 1967 ) .
p. r. page , phys .
c * 72 * , 054312 ( 2005 ) .
a. m. mukhamedzhanov et al . , phys .
c * 81 * , 054314 ( 2010 ) .
a. m. mukhamedzhanov and r. e. tribble , phys . rev .
* c 59 * , 3418 ( 1999 ) .
p. descouvemont and d. baye , rep .
* 73 * , 036301 ( 2010 ) . c. r. gould ,
j. m. joyce , and j. r. boyce , proceedings of conference on nuclear cross sections and technology , n. b. s. publication 421 , washington dc 1975 , p .697 .
d. r. tilley et al .
( unpublished ) ; this is part of the updated version of f. ajzenberg - selove , nucl . phys .
a * 490 * , 1 ( 1988 ) .
|
background : : : the information about the @xmath0li@xmath1he reaction rates of astrophysical interest can be obtained by extrapolating direct data to lower energies , or by indirect methods . the indirect trojan horse method , as well as various r - matrix and polynomial fits to direct data , estimate the electron screening energies much larger than the adiabatic limit .
calculations that include the subthreshold resonance estimate smaller screening energies .
purpose : : : obtain the @xmath0li@xmath2he reaction r - matrix parameters and the bare astrophysical @xmath3 factor for the energies relevant to the stellar plasmas by fitting the r - matrix formulas for the subthreshold resonances to the @xmath3-factor data above 60 kev .
methods : : : the bare @xmath3 factor is calculated using the single- and the two - level r - matrix formulas for the closest to the threshold @xmath4 and @xmath5 subthreshold states at @xmath6 , @xmath7 and @xmath8 mev .
the electron screening potential @xmath9 is then obtained by fitting it as a single parameter to the low - energy data .
the calculations are also done by fitting @xmath9 simultaneously with other parameters .
results : : : the low - energy @xmath3 factor is dominated by the @xmath5 subthreshold resonance at @xmath6 mev .
the influence of the other two subthreshold states is small .
the resultant electron screening is smaller than the adiabatic value .
the fits that neglect the electron screening above 60 kev produce a significantly smaller electron screening potential .
the calculations show a large ambiguity associated with a choice of the initial channel radius .
conclusions : : : the r - matrix fits do not show a significantly larger @xmath10 than predicted by the atomic physics models .
the r - matrix best fit provides @xmath11 ev and @xmath12 mev b.
|
one of the main reasons to build the large hadron collider ( lhc ) at cern is the determination of the mechanism of electro - weak symmetry breaking ( ewsb ) . in the standard model ( sm ) of elementary particle physics and its extensions incorporating supersymmetry ( susy )
, ewsb occurs through the higgs mechanism , which in turn leads to the existence of one or more higgs particles . within the minimal supersymmetric standard model ( mssm ) , a realisation of susy with minimal particle content and gauge structure ,
the higgs potential conserves charge & parity ( cp ) at tree level @xcite . beyond the trivial order
, several studies have shown that cp invariance of the higgs potential may in principle be broken by radiative corrections @xcite , as the vacuum expectation values ( vevs ) of the two higgs doublets can develop a relative phase @xcite .
this type of cp violation is generally referred to as spontaneous cp violation and it requires a light higgs state as a result of the georgi - pais theorem @xcite , but the possibility of the latter has now essentially been ruled out by experiment @xcite@xcite .
cp violation can also be explicitly induced in the mssm , in much the same way as it is done in the sm , by complex yukawa couplings of the higgs bosons to sfermions .
there are several new parameters in the susy theory , that are absent in the sm , which could well be complex and thus possess cp - violating phases . however , the cp - violating phases associated with the sfermions of the first and , to a lesser extent , second generations are severely constrained by bounds on the electric dipole moments ( edms ) of the electron , neutron and muon . nonetheless , there have been several suggestions @xcite@xcite to evade these constraints without suppressing the cp - violating phases . by building on the results of refs .
@xcite ( for the production ) and @xcite@xcite ( for the decay ) see also refs .
@xcite@xcite we will look here at the lhc phenomenology of the @xmath0 process ( where @xmath1 labels the lightest neutral higgs state of the cp - violating mssm ) , which involves the ( leading ) direct effects of cp violation through couplings of the @xmath3 ( @xmath4 corresponding to the three neutral higgs bosons ) to sparticles in the loops as well as the ( sub - leading ) indirect effects through scalar - pseudoscalar mixing yielding the cp - mixed state @xmath3 . ( see ref .
@xcite for some preliminary accounts in this respect . ) all sources of cp violation are reviewed in the next section , where we also lay down our nomenclature and conventions . in the following section we present our results ,
then we conclude .
explicit cp violation arises in the higgs sector of the mssm when various related couplings become complex .
one consequence is that the physical higgs bosons are no more cp eigenstates , but a mixture of these @xcite@xcite .
one may then look at the production and decay of the lightest of the physical higgs particles , hereafter labeled @xmath1 .
cp - violating effects in the combined production and decay process enter through complex @xmath3-@xmath5-@xmath6 couplings at production and decay level plus mixing in the propagator ( @xmath5 represents a sfermion ) .
the nature of these couplings are presented in the appendix .
the leading contribution to higgs production in gluon fusion is at the one - loop level .
similarly , higgs decay into a photon pair also occurs at one - loop level .
these loops contain besides sm and charged higgs states also supersymmetric particles ( @xmath7 and @xmath8 in the case of production while also @xmath9 and @xmath10 in the case of decay ) where the phases of the susy parameters enter . hence cp - conserving and cp - violating effects enter at
_ the same perturbative order _ in the cross section for @xmath11 , so that the latter is an ideal laboratory to pursue studies of the complex ( or otherwise ) nature of the soft susy breaking parameters concerned .
in contrast , notice that cp - violating effects through mixing in the propagator enter only at higher order , through self - energies , as there is already a tree - level cp - conserving contribution to the propagator ( and @xmath12 ) mass is subject to similar loop effects , though @xcite has already shown that the consequent effects are marginal ( see the left - hand side of fig . 4 therein ) ,
so that they are included here but not dwelt upon for long . ] .
the dynamics of cp - violating effects in the production and decay stages are rather similar , given that the same diagrammatic topologies are involved ( see fig .
[ fig : feyn ] ) , in particular , as shown in our previous work @xcite@xcite , we expect a strong impact of a light stop quark in some regions of the parameter space .
the propagator is considered in the following way .
a higgs particle , @xmath3 , produced through gluon fusion , can be converted into another mass eigenstate , @xmath13 , through interaction of fermion or gauge boson loops and their supersymmetric counterparts ( see fig .
[ fig : feyn ] ) .
therefore , in the following , when talking about results for the @xmath1 , we consider the production of any of @xmath14 , which , while propagating , converts into @xmath1 . on the one hand , the proper procedure would be to sum over all three higgs bosons , both in production and decay , and then integrate over an appropriate @xmath15 invariant mass bin , the size of which should be determined by the experimental di - photon mass resolution . on the other hand , in the regions of parameter space that we will cover with this study
it is rarely the case that two ( or more ) higgs boson masses are close together ( within the resolution ) , i.e. , degenerate , where truly cp violating effects at one - loop level can be large @xcite .
in fact , in most of the parameter space volume considered , an _ isolated _
@xmath1 resonance can be seen by experiment , resolved at the level of about 2 gev ( which is roughly the di - photon mass resolution in atlas @xcite , while in cms it is somewhat better @xcite ) , so that we can treat the decay of the @xmath1 separately .
however , whenever this is not possible , because @xmath16 gev , we consider all the degenerate higgs particles together , as explained in section 3 .
as intimated , the propagator matrix is obtained from the self - energy of the higgs particles computed at one - loop level , where we used the expressions provided by @xcite , which include off - diagonal absorptive parts .
the matrix inversion required is done numerically using the lapack package @xcite .
all relevant couplings and masses are obtained from cpsuperh version 2 @xcite , which takes into account all applicable experimental constraints including the low energy edms .
the cross section of the full process shown in fig .
[ fig : feyn ] is computed numerically .
the multi - dimensional integration is carried out using the cuhre program under the cuba package @xcite . for our collider analysis
we have used the cteq6 parton distribution functions ( pdfs ) @xcite@xcite computed at the factorisation / renormalisation scale @xmath17 .
finally , it should be noted that the results obtained here can not be simply extracted from existing programs , as these account for cp - violating effects in production and decay separately , hence without interconnecting them through the appropriately mixed ( and off - shell ) propagators . concerning the latter , we are aware that the inclusion of off - shell higgs propagators in the diagram of fig . [
fig : feyn ] introduces gauge - violating effects , though it should be appreciated that these are at most of order @xmath18 , where @xmath19 is the total width of the higgs state @xmath3 ( @xmath20 ) . for the mass regimes we are interested in here , particularly for the case @xmath21 , such gauge - violating effects are negligible compared to the typical cp - violating ones that we will illustrate in the remainder of the paper
in order to illustrate the typical effects of cp violation in the mssm , we have considered a few sample parameter space points and studied the effect of , in particular , light sparticles in the loops , chiefly , of top squarks .
we fix the following mssm parameters which play only a minor role in cp violation studies for higgs production and decay : + @xmath22 + we consider the case of all the third generation trilinear couplings being unified into one single quantity , @xmath23 .
all the soft masses are taken to be the same at some unification scale , whose representative value adopted here is 1 tev . when considering the light stop case we take a comparatively light value for @xmath24 gev , which corresponds to a stop mass of around 200 gev , otherwise @xmath25 is set to 1 tev .
we could , alternatively , consider small values for @xmath26 to reach light squarks , but the effects would qualitatively be the same .
so , in the following we keep a fixed value of @xmath27 tev . in the higgs scalar - pseudoscalar mixing the product of @xmath28 is relevant rather than @xmath29 or @xmath23 separately .
as argued in our earlier works , the only phase that is relevant is thus the sum of the phases of @xmath29 and @xmath23 . in our analysis
we have kept @xmath30 and studied the effect of cp violation by varying @xmath31 . regarding the absolute values of @xmath29 or @xmath23 , in our numerical analysis , we have varied these parameters between 1 and 1.5 tev .
@xmath32 is instead varied between 100 and 300 gev .
the mass of the lightest higgs particle is consequently in the range of 50130 gev .
we then analyse cases with different @xmath33 s .
in particular , low @xmath33 values give very small deviations from the corresponding cp - conserving cases , while large @xmath33 values produce significant differences .
also , we take a representative value of @xmath34 to see the effect of the other parameters . in addition , to illustrate the cases of very large @xmath33 points , we present the analysis with @xmath35 whilst keeping all other parameters constant . the main aim of this article is to understand the effect of sparticle couplings in the cp - violating mssm . as mentioned earlier , effects of cp violation
enter the process considered and combine in three different ways , viz .
( i ) physical higgs particles are a mixture of the cp eigenstates , ( ii ) through the sparticle - higgs couplings in the production and decay amplitudes , ( iii ) through the higgs mixing in the propagator .
notice though that ( i ) ( and also ( iii ) to some extent ) is a result of ( ii ) , however , in our specific process , ( i)(iii ) all enter giving ( quantitatively and qualitatively ) different effects , so we have listed them separately . as explained in the introduction , effects of ( i ) and ( ii ) in the case of @xmath36 production and @xmath37 decay were already reported on in detail in previous literature , albeit separately , so our intent here is to combine them .
furthermore , we would like to explore the effects due to ( iii ) in our specific process following the general analysis of ref .
@xcite . in fig .
[ fig : omix - hmass ] we plot the physical higgs mass @xmath38 against @xmath32 for a typical parameter space point with @xmath39 tev , @xmath40 tev and @xmath34 , considering three different values of @xmath41 . here
, one can appreciate that the cp - conserving case of @xmath42 does not present any @xmath38 point significantly degenerate with either @xmath43 or @xmath44 for the entire range of @xmath45 gev considered , as the mass difference @xmath46 is always larger than 2 gev .
in fact , in all the cases we considered , in the cp - conserving case @xmath1 , which is the lightest of the physical higgs particles , is a pseudoscalar at low @xmath32 values , which turns to a scalar at large @xmath32 values .
the transition happens around @xmath47 gev , where it is possible for both the physical scalar particles to have masses such that @xmath48 gev .
for the particular @xmath29 , @xmath23 and @xmath33 values considered in fig .
[ fig : omix - hmass ] , this is not the case . away from this mixing region " , the two scalars are well separated in mass . when cp is violated , scalar - pseudoscalar mixing is possible for a large range of @xmath32 values as exemplified in tab .
[ table : mh ] , where @xmath32 and @xmath49 , are tabulated .
this is done for the same set of parameter values as in the case of fig .
[ fig : omix - hmass ] .
.masses of the physical higgs bosons along with the input parameter @xmath32 for @xmath34 , @xmath40 tev , @xmath39 tev and @xmath50 tev in the degenerate case , where @xmath51 gev .
all masses are in gev .
[ cols="^ , > , > , > , > , > , > " , ] fig .
[ fig : prop ] shows the typical effects of cp violation on the cross section through the propagator . in order to understand the latter , we analyse the @xmath1 production , propagation and decay process in the following way .
as intimated already , when the propagator is computed at one - loop level , the considered process could go through a production of any of the three higgs flavours eventually converting into the @xmath1 state . to understand the effect of higgs mixing in the propagator
, we plot the contribution due to the diagonal elements in the propagator matrix and the contribution due to the off - diagonal elements separately , along with the total contribution .
the off - diagonal cp - violating terms contribute around 1% near the degenerate mass region ( @xmath52 gev ) and much less when the higgs masses are distinguishably different .
this effect is seen in all the parameter space points considered in fact , therefore , it is a general feature of the process . in the degenerate region , where two higgs bosons are indistinguishable
, their contributions to the total cross section will have to be added at the amplitude level , and the effect is expected to be distinct from the case of narrow width approximation ( nwa ) . in fig .
[ fig : bw ] ( left ) we plot the cross section of our process for the two cases in the three situations with @xmath42 , @xmath53 and @xmath54 , keeping all other parameters fixed . in this particular case ,
@xmath1 and @xmath55 are degenerate at low @xmath32 values when @xmath53 and @xmath54 . in fig .
[ fig : bw ] ( right ) we plot the percentage deviation of cross section from its value obtained in the nwa , defined as @xmath56 the deviation is between 10 to 30% for @xmath57 , thereby confirming that a production times decay calculation in nwa , neglecting amplitude level superposition in the degenerate case , would lead here to somewhat incorrect results .
such large difference is not expected to arise from higher order corrections . in summary , we would like to point out that it was _ a priori _ not obvious what the effect of propagator mixing would have been , how cp - violating effects in production and decay canceled or supported each other and how the proper accounting for the degenerate mass cases would have impacted the final results .
our analysis makes it clear though that , while the propagator mixing does not affect the considered process in a significant way , the other two work in tandem to make the overall effect significant .
while this statement has been illustrated for some sample parameter space points , we have verified that the above conclusions remain the same for parameter regions with large values of @xmath33 , @xmath29 and @xmath23 , where the cp - violating effects are prominent . finally , as fig .
[ fig : bw ] incorporates cp - violating effects due to the change in the higgs couplings with supersymmetric particles , it is clear that it is the latter which accounts for essentially all the differences between the cp - conserving and cp - violating results , throughout . in this connection , it is of particular importance to investigate the effect of the ( dominant ) higgs - stop - stop couplings with varying stop mass , as illustrated in the discussion below . in fig .
[ fig : tb20a1mu1phimu ] we plot the full cross section for @xmath2 against @xmath32 means the lightest physical higgs boson if it is experimentally distinguishable from the other two higgs bosons .
whenever the lightest one is instead degenerate with one or both of the other two higgs bosons ( in fact , only with @xmath55 for our purposes ) , effects of all the degenerate states are considered at the amplitude level . ] .
we have considered @xmath39 tev , @xmath58 tev and @xmath34 .
there is appreciable variation of the cross section with @xmath31 . comparing the two cases of light and heavy stops ,
it is clear that the effect of the higgs - stop - stop coupling is significant .
indeed , this was also noticed when we studied the di - photon decay @xcite
. figs .
[ fig : tb20a1.5mu1phimu ] and [ fig : tb20a1mu1.5phimu ] illustrate similar studies with @xmath59 , @xmath39 @xmath60 and @xmath58 tev , @xmath61 tev , respectively , in particular showing how significantly the result depends on @xmath23 and @xmath29 . unlike the case of fig .
[ fig : tb20a1mu1phimu ] , where there is only a quantitative difference between the two cases of light and heavy stop , in fig .
[ fig : tb20a1.5mu1phimu ] we see that there is also a qualitative difference between the two cases ( i.e. , in the shape of the cross section curve ) . in contrast ,
[ fig : tb20a1mu1.5phimu ] does only show a quantitative change , yet this simply has to do with the particular values of @xmath29 and @xmath23 considered . in general , we may expect the difference between the cp - conserving and cp - violating cases to depend quite sensitively on @xmath23 and @xmath29 .
[ fig : tb50a1mu1phimu ] shows the cross section for @xmath62 50 with @xmath58 tev and @xmath39 tev . as in the case of a di - photon decay considered independent of the production , here too ( i.e. , for the full process ) cases with small @xmath63 with @xmath58 tev and @xmath39 tev are not very sensitive to cp - violating phases .
we may note that the couplings @xmath64 , eq .
( [ eqn : hisfsf ] ) , as well as other higgs couplings , have complicated relations with various susy parameters , and therefore it is difficult to understand all the features of the cross section in full detail .
nonetheless we have attempted doing so .
we can certainly say that the minimum in the cross section plots ( figs .
48 ) corresponds to the cross - over region in fig .
2 . besides , in all plots presented , the cross section rises ( the more the larger @xmath65 , so long that it is above 145 gev or so ) with @xmath31 except in the right plot of fig . 6 ( similarly to the top - right frame of fig . 11 in @xcite ) and this can be traced back to the shape of the @xmath64 coupling itself .
in fact , we generally recover in the case of the full process @xmath66 the trends already seen in ref .
@xcite for the di - photon decay alone , signifying that cp - violating effects at production level are not offsetting those affecting the decay ( as those induced by propagation are negligible ) , in fact both are at times cooperating to enhance the overall corrections .
also , one can often appreciate sudden variations in the di - photon yield due to indirect width effects .
for example , the small hump observed in fig .
[ fig : tb20a1mu1.5phimu ] and in fig .
[ fig : tb50a1mu1phimu ] for @xmath67 around @xmath68 gev is an indirect effect on the @xmath69 due to a dip in the total width of @xmath1 corresponding to these parameter values .
the dip in the total width itself arises because @xmath1 is purely @xmath70 type around here , and therefore , does not couple to the charged leptons and bottom quarks .
furthermore , it should also be recalled that a common feature to all graphs is that the @xmath1 is mostly cp - mixed for small @xmath65 values but after the cross over around @xmath71 gev it is mostly cp - even ( see fig . 5 of ref .
@xcite ) , this also impinging on the shape of the overall cross section . finally , while in many cases the entire allowed range of the cross section can be spanned by simply considering both signs of @xmath29 ( the two cp - conserving limits , @xmath42 and @xmath72 ) , where the results for cp - violating scenarios lie between these two cases , in many of the instances studied , owing to the structure of the cp - violating couplings between squark and higgs states ( which involve two terms with , at times , contrasting signs ) , this is generally not the case when @xmath73 gev and also in some instances when @xmath74 gev ( see , e.g. , the right hand side of fig
. 6 ) .
in addition to the stop mass dependence of the @xmath66 cross section , we have also studied how the latter varies with the masses of the other ( s)particles entering the loops .
however , in line with the results of refs .
@xcite@xcite for the case of the @xmath75 decay , we have found that their impact is largely negligible here too , no matter the value of the cp - violating phases .
nonetheless , one last aspect ought to be explored in connection with the @xmath76 dependence : how the value of the latter is itself affected by the cp - violating phases . in fig .
[ fig : stopmass ] we plot the variation of the lightest stop mass with the phase of @xmath29 for different parameter space points considered in the earlier plots .
for example , if @xmath34 , when @xmath61 tev and @xmath58 tev , the mass @xmath76 lies within @xmath77 of its average value whereas , for @xmath39 tev and @xmath40 tev , it lies within @xmath78 of it . for @xmath35
the spread is even narrower ( @xmath79 ) while for @xmath80 it is larger ( @xmath81 ) .
notice though that these variations are typically smaller than the experimental resolution expected at the lhc in measuring @xmath76 , which are of order 20% or so @xcite .
therefore , an intriguing prospect appears , that the discovery of a light mssm higgs boson ( with mass below 130 gev or so ) at the lhc may eventually enable one to disentangle the cp - violating case from the cp - conserving one , so long that the relevant susy parameters entering @xmath82 are measured in the same process or elsewhere , in particular @xmath38 ( or alternatively @xmath65 , recall our fig . 2 here and/or see ref .
@xcite for the relation between the two ) and @xmath76 .
this is not phenomenologically inconceivable , as , on the one hand , the @xmath15 final state is ideally suited to measure the higgs mass ( or indeed resolve superpositions of higgs states ) given the high di - photon mass resolution , and , on the other hand , this higgs detection mode requires a very high luminosity , unlike the discovery of those sparticles ( and the measurement of their masses and couplings ) that enter the loops , chiefly top squarks .
finally , notice that it is from this point of view that we decided to neglect studying here the impact of both reducible and irreducible backgrounds onto our signal process .
( the former typically include @xmath83 jet and di - jet production , with jets mis - tagged as photons , whereas the latter involve the tree - level @xmath84 and one - loop @xmath85 [ via box diagrams involving quarks and squarks ] subprocesses . ) on the one hand , any cp - violating effect proportional to squark flavour mixing mass terms coming from non - resonant diagrams , such as the gluon - gluon - gamma - gamma box diagram , is expected to be very small owing to the stringent experimental constraints on rare cp and flavour changing neutral current ( fcnc ) processes .
on the other hand , since the background processes including cp - violating effects at higher order do not produce any kinematic structure around the presumed @xmath1 resonance , they can accurately be measured off peak and smoothly interpolated beneath it .
we have shown the conspicuous effects of cp violation onto the lhc cross section for @xmath2 in the mssm , as the overall production and decay rate varies over an order or so of magnitude , depending on the value of @xmath31 ( or , alternatively , @xmath86 ) .
varying the mass of the top squark from about 1 tev to around 250 gev has also a strong quantitative impact on the cross section , possibly changing its dependence on the phases also qualitatively .
in fact , a significant dependence of the overall cross section on @xmath31 when @xmath1 is not a cp - mixed state ( as seen from all the figures presented for large @xmath65 values ) clearly exemplifies the importance of the higgs - stop - stop coupling when cp violation is considered .
altogether then , it is clear that the prime detection channel of a light mssm higgs boson at the lhc is also a prime candidate to prove the existence of explicit cp violation in minimal susy .
in fact , the @xmath31 dependence of our results is strongest when the cross section is away from the minima , typically for @xmath65 larger than 150 gev or so .
s. munir s research is sponsored by dgapa , unam , and conacyt , mexico , project no . 82291-f .
+ @xmath88 where @xmath89 is the weak gauge coupling , @xmath90 for @xmath91 and @xmath92 for @xmath93 . here
@xmath94 corresponds to @xmath95 , @xmath96 , @xmath97 quarks , @xmath98 corresponds to @xmath99 , @xmath100 , @xmath101 quarks and @xmath102 stands for @xmath103 , @xmath104 , @xmath105 leptons .
the matrix @xmath106 is a real @xmath107 matrix which takes the higgs fields from their electroweak eigenstates to the physical mass eigenstates : @xmath108 + @xmath109 the couplings @xmath116 in the @xmath117 basis are given by @xmath118v\sin\beta&-\frac{1}{\sqrt{2}}h^*_ta_t^*\\ -\frac{1}{\sqrt{2}}h_ta_t & \left[-|h_t|^2+\frac{1}{3}g'^2\right]v\sin\beta \end{array}\right).\end{aligned}\ ] ] here @xmath119 is the top yukawa coupling , @xmath89 and @xmath120 .
expressions for the couplings @xmath121 in the case of other third generation sfermions ( which are of relevance to our study ) are given in , for example , @xcite .
for a review , see : j. f. gunion , h.h e. haber , g. kane and s. dawson , _ the higgs hunter s guide _ ( addison - wesley , reading , ma , 1990 ) . n. maekawa , phys .
b * 282 * , 387 ( 1992 ) .
a. pilaftsis , phys .
b * 435 * , 88 ( 1998 ) .
h. georgi and a. pais , phys .
d * 10 * , 1246 ( 1974 ) .
a. pomarol , phys .
b * 287 * , 331 ( 1992 ) .
n. haba , phys .
b * 398 * , 305 ( 1997 ) .
o. c. w. kong and f. l. w. lin , phys .
b * 419 * , 217 ( 1998 ) .
nath , phys .
lett . * 66 * , 2565 ( 1991 ) .
y. kizukuri and n. oshimo , phys .
d * 45 * , 1806 ( 1992 ) ; _ ibid .
d * 46 * , 3025 ( 1992 ) .
s. abel , s. khalil and o. lebedev , nucl .
b * 606 * , 151 ( 2001 ) .
a. bartl , w. majerotto , w. porod and d. wyler , phys .
d * 68 * , 053005 ( 2003 ) .
k. a. olive , m. pospelov , a. ritz and y. santoso , phys . rev .
d * 72 * , 075001 ( 2005 ) .
s. abel and o. lebedev , jhep * 0601 * , 133 ( 2006 ) .
s. dimopoulos and g. f. giudice , phys .
b * 357 * , 573 ( 1995 ) .
a. g. cohen , d. b. kaplan and a. e. nelson , phys .
b * 388 * , 588 ( 1996 ) .
a. pomarol and d. tommasini , nucl .
b * 466 * , 3 ( 1996 ) .
s. yaser ayazi and y. farzan , phys .
d * 74 * , 055008 ( 2006 ) .
t. ibrahim and p. nath , phys .
b * 418 * , 98 ( 1998 ) ; phys . rev .
d * 57 * , 478 ( 1998 ) [ erratum - ibid .
d * 58 * , 019901 ( 1998 errat , d60,079903.1999 errat , d60,119901.1999 ) ] . t. falk and k. a. olive , phys .
b * 439 * , 71 ( 1998 ) .
m. brhlik , g. j. good and g. l. kane , phys .
d * 59 * , 115004 ( 1999 ) .
s. pokorski , j. rosiek and c. a. savoy , nucl .
b * 570 * , 81 ( 2000 ) .
j. r. ellis , j. s. lee and a. pilaftsis , jhep * 0810 * , 049 ( 2008 ) .
k. cheung , o. c. w. kong and j. s. lee , jhep * 0906 * , 020 ( 2009 ) .
a. dedes and s. moretti , phys .
lett . * 84 * , 22 ( 2000 ) .
a. dedes and s. moretti , nucl
. phys .
b * 576 * , 29 ( 2000 ) .
s. moretti , s. munir , p. poulose , phys .
b * 649 * , 206 ( 2007 ) .
s. hesselbach , s. moretti , s. munir , p. poulose , eur .
j. c * 54 * , 129 ( 2008 ) .
s. hesselbach , s. moretti , s. munir , p. poulose , j. phys .
ser . * 110 * , 072017 ( 2008 ) .
s. hesselbach , s. moretti , s. munir , p. poulose , arxiv:0710.4923 [ hep - ph ] .
a. dedes and s. moretti , _
prepared for workshop on physics at tev colliders , les houches , france , 21 may - 1 jun 2001 .
_ s. y. choi and j. s. lee , phys . rev .
d * 61 * , 115002 ( 2000 ) .
s. y. choi , k. hagiwara and j. s. lee , phys .
b * 529 * , 212 ( 2002 ) .
j. r. ellis , j. s. lee and a. pilaftsis , phys .
d * 70 * , 075010 ( 2004 ) ; mod .
a * 21 * , 1405 ( 2006 ) .
s. hesselbach , s. moretti , s. munir and p. poulose , arxiv:0910.0230 [ hep - ph ] .
a. pilaftsis and c. e. m. wagner , nucl .
b * 553 * , 3 ( 1999 ) .
m. s. carena , j. r. ellis , a. pilaftsis and c. e. m. wagner , nucl .
b * 586 * , 92 ( 2000 ) .
s. heinemeyer , int .
j. mod .
a * 21 * , 2659 ( 2006 ) .
m. frank , t. hahn , s. heinemeyer , w. hollik , h. rzehak and g. weiglein , jhep * 0702 * , 047 ( 2007 ) .
atlas collaboration , atlas detector and physics performance : technical design report , 2 , cern - lhcc-99 - 015 ( 1999 ) .
cms collaboration , cms physics : technical design report v.2 : physics performance , cern - lhcc-2006 - 021 ( 2006 ) .
see http://www.netlib.org / lapack/. j. s. lee , m. carena , j. ellis , a. pilaftsis and c. e. m. wagner , comput .
commun . *
180 * , 312 ( 2009 ) .
t. hahn , comput .
commun . * 168 * , 78 ( 2005 ) .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. m. nadolsky , w. k. tung , jhep * 0207 * , 012 ( 2002 ) .
d. stump , j. huston , j. pumplin , w. k. tung , h. l. lai , s. kuhlmann , j. f. owens , jhep * 0310 * , 046 ( 2003 ) .
j. f. owens _ et al .
_ , phys .
d * 75 * , 054030 ( 2007 ) .
h. l. lai , p. m. nadolsky , j. pumplin , d. stump , w. k. tung , c. p. yuan , jhep * 0704 * , 089 ( 2007 ) .
|
we prove the strong sensitivity of the @xmath0 cross section at the large hadron collider on the explicitly cp - violating phases of the minimal supersymmetric standard model , where @xmath1 is the lightest supersymmetric higgs boson .
-0.125 in -0.125 in 6.5 in 8.75 in dcpt-09 - 26 + ippp-09 - 13 + shep-08 - 36 + revised * explicit cp violation in the mssm + through @xmath2 * pacs : 14.80.da , 14.80.ec , 11.30.er
|
the am canum venaticorum ( am cvn ) variables are a class of he - rich objects with variability periods of @xmath11566 minutes .
various lines of evidence ( see 1 of * ? ? ?
* for a brief summary ) indicate these periods are orbital in origin and that ongoing mass - transfer from a low - mass , essentially pure - he donor onto a white dwarf ( wd ) accretor is taking place .
thus , am cvn systems form the wd - accreting class of so - called ultracompact binaries : interacting stellar binaries with orbital periods , @xmath0 , below the minimum @xmath0 attainable by h - dominated objects ( @xmath117080 minutes , see , e.g. , * ? ? ?
* ) . at such short @xmath0 ,
the binary evolution is driven by orbital angular momentum losses due to gravity - wave ( gw ) emission .
the am cvn objects represent an extreme end - product of stellar - binary evolution and at least some of them are examples of wd - wd binaries that survived the transition from a detached phase of gw - driven in - spiral to their current state of stable mass - transfer .
there are significant uncertainties concerning aspects of both the prior binary evolution ( most importantly the outcome of common - envelope events ; * ? ? ?
* ) and the outcomes of wd - wd binaries initiating mass transfer .
the characteristics of the am cvn population can thus provide insights into the physics important to both these phases of their prior evolution .
it is expected that space - based gw interferometers , such as the proposed _
lisa _ mission , will detect essentially the entire galactic population of am cvns with @xmath12 minutes @xcite , providing an observational picture of unprecedented detail .
future prospects for strong constraints from the am cvn population on our theories of binary formation and evolution are , thus , very bright .
in the meantime , advances from electromagnetic observations and developing theory will begin addressing these same questions .
there have been three distinct binary evolution channels proposed to form am cvn systems .
two of these formation channels involve a series of two common - envelope ( ce ) events which bring the remnant cores of a main - sequence ( ms ) binary close enough for gw - emission to drive them into contact , initiating the am cvn phase .
these channels are distinguished by the state of the proto - donor prior to contact @xcite . in what we ll call the he - star channel
, the proto - donor is able to ignite he prior to contact . in the other , which we ll refer to as the wd channel
, the donor is partially - to - very degenerate @xcite and has not undergone any he burning .
the third channel involves an evolved ms star with a wd companion starting stable mass transfer just before it evolves up the red - giant branch @xcite .
the star s core is dominated by he , allowing the binary to evolve to ultracompact configurations .
we ll refer to this third channel as the evolved - ms channel . in mass transferring binaries , the donor s structural response to mass loss plays a central role in determining the binary s evolution ( see , e.g. , * ? ? ?
the extent to which the donors from each formation channel have been modelled varies considerably .
he - star channel donors have only been modelled by a fit @xcite to a single relevant binary evolution calculation by @xcite .
the wd channel donors have been considered more extensively .
@xcite modelled these donors as fully - degenerate wds obeying the zero - temperature he wd mass - radius ( @xmath13-@xmath14 ) relation .
later , @xcite , using donor structural models developed for ultracompact low - mass x - ray binaries @xcite , modelled the evolution of non - zero entropy donors from the wd - channel .
@xcite has presented a detailed study of the evolved - ms star channel . in this paper , we focus on the wd channel systems .
the @xcite study had several limitations : they assumed the donors to be fully convective and to evolve adiabatically under mass loss due to the very large mass transfer rates , @xmath15 , produced in am cvns . with these assumptions ,
the evolution is completely determined by the total mass of the two components and the donor s specific entropy , which determines the @xmath13-@xmath14 relation the donor follows and , amongst other system parameters , the binary s @xmath16 evolution .
the inclusion of non - zero entropy donors in this population resulted in shifting systems to larger @xmath15 at fixed @xmath0 and allowed systems to evolve out to longer @xmath0 as compared to a population of zero - temperature donors .
this provided an observational diagnostic of an am cvn system s initial conditions .
the @xcite calculation also indicated that wd channel systems with hot donors and he - star channel systems would not be distinguishable based on @xmath15 determinations alone .
since the entropy structure of the donor determines its response to mass loss ( see the discussion in [ sec : donorev_prer2min ] and appendix [ app : r_response ] ) , assuming a fully convective structure will overestimate the binary s @xmath0 evolution rate if the donor s true structure in not adiabatic .
further , while assuming the donor responds adiabatically to mass - loss is certainly valid at early stages of the am cvn phase when @xmath15 is very large , this assumption may break down as the binary evolves outward in @xmath0 to lower @xmath15 . to rectify these shortcomings and
provide predictions for the donor s luminosity and effective temperature , we undertook a study to model the donors within the context of a full stellar structural calculation coupled to the evolution of the am cvn binary .
in particular , we sought to determine the donor s observational signatures as a function of initial system parameters and @xmath0 , as well as to determine how a complete treatment of the donor s structure alters the conclusions of earlier studies .
we also sought to model accurately the critical phase of mass transfer turn - on , when the donor is coming into contact and a transition in the @xmath0-evolution from gw - dominated in - spiral to the mass - transfer dominated expansion occurs .
this phase of wd channel system evolution has not been modelled in any detail prior to this study .
to carry out the computations , we developed a new stellar evolution code to perform the coupled donor / binary calculations .
a brief description of this code , detailing our numerical modelling of the donor and binary is provided in [ sec : calcdetails ] . in [
sec : initial_models ] , we describe the procedure for determining the range of initial donor degeneracy and donor structural models used for our am cvn phase calculations .
we present the results of our calculations in [ sec : ev_general ] and discuss these results and several of their applications in [ sec : discussion ] .
finally , we summarize in [ sec : summary ] .
to achieve a necessary level of flexibility in terms of input physics and defining system of equations for this and related projects , we implemented a stellar evolution code capable of handling most 1d stellar evolution calculations expressed as boundary - value problems .
this code is written in c++ and is comprised of a suite of abstract classes defining the necessary components of a stellar evolution calculation . at its core is a generalized 1d relaxation - method boundary - value problem ( bvp ) solver ( see , e.g , 17.3 of * ? ? ?
* ) that utilizes the sparse - matrix solver umfpack to perform the necessary matrix inversions .
the defining system of ordinary differential equations ( odes ) and necessary input physics are implemented as separate abstract classes , providing the desired flexibility . utilizing this new code
, we solve for both the donor s structure and binary parameters within a single set of relaxation iterations at each time step .
below we detail our numerical treatment for both components of our system and describe the input physics . for the donor , we solve the standard set of 1d hydrostatic stellar structure equations augmented by an automatic mesh allocation algorithm @xcite on a 200-point non - lagrangian mesh : @xmath17 here @xmath18 , @xmath19 , @xmath20 , @xmath21 , and @xmath22 , are the density , temperature , pressure , radius , and luminosity at each mesh point , @xmath23 is the mass interior to each mesh point , @xmath24 is time , @xmath25 is the local nuclear energy generation rate ( which we set to zero throughout this work ) , and @xmath26 is the specific heat at constant pressure . in equation ( [ eq : dqdm])which determines the mesh allocation@xmath27 and @xmath28 are the pressures at the stellar centre and exterior point , respectively , @xmath29 are constants that weight the importance of pressure and mass gradients in determining the mesh spacing , and @xmath30 is an overall normalization constant determined implicitly at each time step . the quantity @xmath31 , where @xmath32 is the specific entropy , and @xmath33 describes the actual run of @xmath19 with @xmath20 within the donor .
we take @xmath34 when @xmath35 , where @xmath36 and @xmath37 is the rosseland mean opacity .
otherwise we determine @xmath38 using mixing length theory with the mixing length set equal to one pressure scale height .
the central boundary conditions are determined by the standard central expansions of the stellar structure equations . at the outer boundary
, we equate the values of the iterated quantities at the outermost mesh point , @xmath39 , and @xmath40 with the corresponding values at the photosphere ( i.e. , we do not calculate a separate atmosphere model ) .
grey atmospheres and the eddington approximation are assumed .
the photospheric pressure is therefore given by : @xmath41 where @xmath42 , @xmath43 is the photospheric density , and @xmath5 is the donor s effective temperature .
this gives the matching conditions @xmath44 , @xmath45 , @xmath46 , and @xmath47 , and @xmath48 .
here @xmath49 , @xmath50 and @xmath4 are the donor s total mass , radius , and luminosity , with @xmath51
. to produce a smoothly varying outer boundary condition , equation ( [ eq : photosphere ] ) is not solved directly each time photospheric values are required . instead
, a @xmath52 grid of @xmath53 points is constructed that straddle the donor s current photospheric conditions .
we calculate @xmath54 from equation ( [ eq : photosphere ] ) at these values and use these to determine @xmath54 by cubic - spline interpolations during the iterations . if a point outside of the current grid is needed , the grid is re - centred and new @xmath54 values calculated .
the @xmath5 and @xmath55 grids have spacings @xmath56 , @xmath57 . to specify the binary system , we add the following three equations to the above set of odes : @xmath58 and @xmath59 \label{eq : dadt } \,,\end{gathered}\ ] ] where @xmath60 is the accretors mass
, @xmath61 is the orbital separation , @xmath62 is the orbital angular momentum .
the quantities @xmath63 , @xmath64 specify the mass evolution rate of the accretor and donor ; both are , in general , functions of the binary s and , possibly , the donor s state .
we express the @xmath61 evolution in the form of equation ( [ eq : dadt ] ) to allow for this generality .
_ however , here , we consider conservative mass transfer , setting @xmath65 always_. we consider only gravity wave emission @xcite contributions to the @xmath62 evolution , ignoring the possibility of accretor spin - up in the @xmath62 evolution .
this latter effect may play a significant role in determining whether an am cvn binary survives the onset of contact @xcite .
this omission is justified since we are interested in understanding the parameter space potentially allowed to am cvn systems , not in a detailed characterization of the actual , realized population
. this latter question will be the subject of future studies . to determine @xmath64 we utilize the prescriptions of @xcite and @xcite . specifically ,
when the donor s roche radius , @xmath66 , here calculated as @xcite @xmath67 ( valid for @xmath68 ) is larger than @xmath50 , the mass loss is modelled as an isothermal flow of a classical gas with a rate @xmath69 where @xmath70 is the mass transfer rate off the donor when @xmath71 and depends on the surface properties @xcite , @xmath72 is the pressure scale height at the photosphere , and @xmath73 .
when @xmath74 , we model the sub - photospheric contributions with an adiabatic flow @xcite . to do so , we rewrite equation ( a17 ) of @xcite in a more general form to allow for degeneracy or non - ideality : @xmath75 where @xmath76 and @xmath77 are defined in @xcite and @xmath78 .
we determine @xmath15 implicitly at each timestep , a method that greatly improves the numerical stability @xcite . for the donor s composition ,
we assume a fixed , homogeneous mixture with he , c , n , and o mass fractions of 0.981 , @xmath79 , 0.0122 , and @xmath80 ( typical of the core composition of solar metallicity stars at the base of the red - giant branch that have undergone cno - cycle burning ; * ? ? ?
* ; * ? ? ?
the remaining metals are assumed to have a solar abundance pattern .
we use opal radiative opacities @xcite calculated for this mixture for temperatures above @xmath81 k. below @xmath81 k , we use @xcite opacities calculated _ excluding _ contributions due to grains ( which form beginning at @xmath82 k ) . where conductive opacities are relevant , we utilize those of @xcite . as
metals make negligible contributions to the donor s eos @xcite , the pure - he eos of @xcite supplemented by radiation contributions are used .
this eos is calculated using a free - energy minimization model and provides the thermodynamic quantities of a he fluid over a wide range of density and temperature including , in particular , the regimes where he undergoes thermal and pressure ionization state transitions .
as shown in @xcite , evolution during the post - contact am cvn phase is strongly influenced by the state of the donor at contact . in this section ,
we determine a set of initial donor models that encompass the range of donor properties at contact produced in wd channel am cvn systems and discuss the expected distribution of these properties based on the population synthesis calculation of @xcite . to determine the range of parameters our initial donor models should cover , we utilize data from the @xcite population synthesis model and the methodology of @xcite .
specifically , for each wd channel system in the population synthesis model that is expected to survive initial contact ( see * ? ? ?
* ) , we determine the proto - donor s core conditions at the beginning of the second ce using existing stellar evolution calculations performed with the ez code @xcite for the @xcite study . during the rapid ce phase ,
the proto - donor s central degeneracy does nt change .
so we utilize the central degeneracy parameter@xmath83 , where @xmath84 , @xmath85 , @xmath86 are the central density , temperature , and electron fermi energy to map between the ez stellar models pre - ce and a set of post - ce isolated he wd models calculated using with our own stellar evolution code .
once we determine this post - ce donor model , for each system we evolve the donor s radius and the orbital separation from this post - ce state to determine when gw emission drives the binary back into contact .
the population synthesis data provides the post - ce orbital separation , @xmath87 , and the initial accretor and donor masses , @xmath88 and @xmath89 .
we use these quantities and our single he - star tracks to calculate , self - consistently , the time elapsed in the post - ce phase , @xmath90 , before @xmath71 .
this determines the donor s contact radius and @xmath91 ( which we denote by @xmath92 and @xmath93 ) .
when taken over the set of population synthesis data , this procedure allows us to determine the range of @xmath93 , @xmath89 , and @xmath88 that occurs in the wd channel systems .
before discussing the distribution of donor contact parameters obtained with this procedure , we mention one complication we encountered . in some systems , the standard ce - evolution prescription ( see , e.g. , * ? ? ?
* ) predicts post - ce conditions where _ the donor is already in roche contact_. this is due to some systems having sufficiently hot proto - donors with @xmath50 significantly greater than the zero - temperature configuration . it is unclear whether such systems will cleanly exit the ce or simply be driven to a prompt merger .
a realistic determination of this question is beyond the scope of this paper . instead , in order to examine how this issue could influence the distribution of donor contact parameters , we considered two different , _ rough _ criteria for determining which systems survive the ce .
in one , we removed all such systems from the population . in the second , we looked at the evolution of @xmath50 ( based on the _ isolated _ he wd structures ) and @xmath66 to determine if @xmath94 was ever satisfied before the system was driven to @xmath95 due to gw emission .
this could happen if the proto - donor is able to contract more rapidly than gw emission decreases @xmath66 .
we found that the choice of criteria effected mainly the distribution of @xmath93 at fixed @xmath89 for systems with @xmath96 . taking the second criteria ,
about 50% of systems with @xmath97 and 25% of those with @xmath98 do not exit the ce by this rough criteria , where as only 4% of those with @xmath99 are affected .
more importantly , the range of @xmath93 in the overall population is not affected by these border line systems , since donors with @xmath100 exhibit the full range of @xmath93 and these more massive donors all appear to exit the ce cleanly .
this general conclusion also applies to concerns about donors with @xmath101 .
the progenitors of these donors will not have developed a distinct core / envelope structure at the start of the ce , making it unlikely that they will survive the ce - event @xcite .
thus , excluding such systems from consideration will also not influence the @xmath93 range expected in this population .
the results of the above exercise produces a population with @xmath102 and @xmath103 ( i.e. , donors are mildly to extremely degenerate ) .
our determination of the donor s @xmath104 , @xmath85 at contact for each system in the @xcite population synthesis model , is shown in figure [ fig : contact_rhoctc ] by crosses .
each cross represents the starting point for a number of systems , so the density of crosses does not correspond to the number of systems starting within a given region of this parameter space .
dashed lines show lines of constant @xmath91 , while solid curves show the evolution along our isolated he wd tracks .
we have excluded systems not exiting the ce according to our second criteria above from this plot ; if we were to use the first criteria , the only change here would be the loss of most systems with @xmath105 and @xmath106 .
the solid circles in this figure show the central conditions of the initial donor models we use in subsequent calculations . , @xmath85 at contact for each wd channel am cvn system in the @xcite population synthesis model ( crosses ) . for comparison ,
the solid lines show the @xmath84 , @xmath85 evolution along each of our isolated he - star evolution tracks ( with @xmath49 values of , from bottom to top , 0.10 , 0.15 , 0.20 , 0.25 , 0.30 , and 0.325 @xmath107 ) . lines of constant @xmath1081.0 , 2.0 , 3.0 , and 4.0 are shown by the dashed lines ( top to bottom ) .
the solid circles show the central conditions of the set of initial donor models used to begin our am cvn phase calculations.,title="fig:",width=317 ] we show the distribution in orbital period , @xmath0 , at contact ( integrated over all @xmath89 ) in figure [ fig : contact_porb ] with the solid line . in this plot , the number , @xmath109 , in each bin represent the total number of wd channel am cvn systems in the @xcite population synthesis model that make contact within each @xmath0 range . for comparison
we also show with the dashed line the distribution that results from assuming fully degenerate donors .
we take @xmath0 at contact to equal @xmath110 minutes and calculate the fully degenerate @xmath92 using a fit to the @xcite zero - temperature he mass - radius relation : @xmath111 which is accurate to better than 3% over the range @xmath112 .
the fully degenerate distribution peaks at @xmath113 minutes with a cut - off at 6 minutes . on the other hand ,
roughly 25% of systems within our current calculation make contact at a @xmath114 minutes and 5% at @xmath115 minutes .
additionally , there is a small tail of systems making contact out to @xmath116 minutes ( not shown in figure [ fig : contact_porb ] ) .
if we remove all systems with post - ce @xmath117 , this tail extends out to 22 minutes , but otherwise the contact @xmath0-distribution is almost unchanged .
-distribution as determined by self - consistent treatment of the donor s
cooling before contact ( solid histogram ) compared to assuming all donors are fully degenerate at contact ( dash dotted histogram ) .
the number of systems in each bin gives the total number of current wd channel am cvn systems in the @xcite population synthesis model whose contact @xmath0 fell within each bin .
the solid and dashed curves show the corresponding cumulative fractional distribution of contact @xmath0.,title="fig:",width=317 ] the corresponding @xmath93-distribution is shown by the histogram in figure [ fig : psici_hist ] .
the solid curve in this figure gives the cumulative distribution of @xmath93 values .
noteworthy is the fact that , unlike the contact @xmath0-distribution , which is strongly biased towards short @xmath0 , the @xmath93-distribution is approximately flat between @xmath118 and 3.6 .
the difference in the two distributions is due to the dependence of @xmath50 on @xmath93 : @xmath50 varies rapidly with @xmath93 for @xmath119 , but only more slowly for @xmath120 .
thus there is not as strong an a priori theoretical preference for large @xmath93 values in the wd channel population as might be expected from the contact @xmath0-distribution .
the @xmath93-distribution is essentially unchanged when removing all systems with post - ce @xmath117 from consideration .
distribution , integrated over all @xmath89 , normalized to the total number of systems in the population model .
the solid curve provides the cumulative distribution . unlike the contact @xmath0 distribution in figure [ fig : contact_porb ] , which is strongly peaked at shorter periods ,
the @xmath93 distribution is relatively flat between @xmath118 and 3.6 .
, title="fig:",width=317 ]
in order to adequately sample the range of initial donor properties derived in [ sec : initial_models ] in our subsequent calculations , we considered am cvn evolution starting from 36 different initial donor models ( solid circles in fig .
[ fig : contact_rhoctc ] ) with @xmath89 values of 0.10 , 0.15 , 0.20 , 0.25 , 0.30 , and 0.325 @xmath107 and @xmath121 values of 1.1 , 1.5 , 2.0 , 2.5 , 3.0 , and 3.5 . the coupled binary and donor evolution
were calculated for two @xmath88 for each donor model .
these @xmath88 were chosen so that @xmath122 equalled one of three values0.5 , 0.825 , or 1.325 @xmath107 . to present our results , we first discuss the donor s evolution in detail for several representative cases .
we then examine the entire range of our calculations , detailing how and in which evolutionary phases different initial conditions affect the system s parameters .
one of our main results is that the donor s outer boundary condition is dominated by irradiation from the accretion flow .
we also , therefore , present additional calculations including the effects of external irradiation and discuss the impact of this additional physics on the system s evolution .
figure [ fig : r_lagrange_ev ] shows a summary of the donor s evolution in a representative calculation .
the initial conditions there were @xmath123 , @xmath124 , @xmath125 , and @xmath126 .
this model illustrates the range of donors responses to mass loss ; for comparison , figure [ fig : r_lagrange_evb ] shows a similar summary for a more degenerate , @xmath127 , donor . in figure
[ fig : r_lagrange_ev ] , the solid black lines ( except for the top - most ) show the @xmath21 evolution at fixed @xmath23 as a functions of time ( the calculations are started with @xmath128 at an @xmath61 such that @xmath129 ) .
the top - most solid line traces out the @xmath50 evolution .
( black lines , the bottom line shows @xmath130 with higher lines spaced by @xmath131 ; the top - most black lines traces @xmath50 ) , the @xmath21 at which @xmath132 ( dashed red line , below this line @xmath133 ) , and the lower boundary of the outer convective zone ( dashed green line ) .
the initial conditions here are @xmath123 , @xmath134 , @xmath125 , and @xmath126 .
, title="fig:",width=317 ] the other lines in this plot indicate various physical conditions in the donor .
all donor models have , to some extent , an outer convective region ; the dashed - dotted green line indicates the lower radius to which this convective zone penetrates . the dashed red line indicates the @xmath21 at which the local thermal time , @xmath135 , equals the local mass loss time , @xmath136 , where these two quantities are defined as @xmath137 @xmath138 and @xmath139 is the mass exterior to the specified location in the donor
regions below the dashed red line in figure [ fig : r_lagrange_ev ] satisfy @xmath133 .
the evolution in layers where @xmath140 is dominated by the nearly adiabatic advection to lower @xmath20 , while layers where @xmath141 are able to adjust their thermal structure in response to mass loss almost instantaneously .
the results shown in figure [ fig : r_lagrange_ev ] are typical of our calculations in that the evolution there can be divided into three phases . in the first phase , during which @xmath15 grows from zero to its maximum , @xmath50 decreases towards a minimum value in response to mass loss . in fig .
[ fig : r_lagrange_ev ] , this phase lasts until @xmath142 yrs .
the second phase begins as the @xmath50 evolution reverses and the donor begins expanding in response to mass loss .
this expansion phase corresponds to what is normally considered the am cvn phase of evolution . by this point ,
@xmath140 throughout the donor so that the donor responds adiabatically to the mass loss .
note that the donor enters this adiabatic phase well before the @xmath50 evolution reverses ( see below ) . at some point during the expansion phase ,
the @xmath132 line begins moving inward again .
this eventually leads to the start of the third phase , where the donor is able to cool and contract .
this is seen at @xmath143 yrs in fig .
[ fig : r_lagrange_ev ] .
this contraction ends once the donor has shed sufficient entropy to reach its fully degenerate configuration , even as it continues to cool ( evolution beyond @xmath144 yrs in fig .
[ fig : r_lagrange_ev ] ) . , but with a more degenerate donor ; the initial conditions in this system are @xmath145 , @xmath146 , @xmath147 , and @xmath148,title="fig:",width=317 ] the first and third phases of the donor s evolution are qualitatively new results resulting from our more complete treatment of the donor s physics . a more detailed discussion of these phases follows .
once the wd channel am cvn donors make contact , our calculations show they all begin a phase of radius contraction , a result not anticipated by prior modelling ( e.g. , * ? ? ?
* ; * ? ? ?
, we discuss how the donor s structure determines the duration and extent of this initial @xmath50 contraction phase .
the basic picture of the donor s response to mass loss can be seen from figs .
[ fig : r_lagrange_ev ] and [ fig : r_lagrange_evb ] .
as @xmath49 decreases , mass elements move to lower @xmath20 and @xmath18 , producing the expansion in @xmath21 at fixed @xmath23 seen in the evolution of the interior black lines . in appendix
[ app : r_response ] we show that this expansion is most significant near the surface and that the only contribution to @xmath50 contraction comes from the surface term in equation ( [ eq : drdmint ] ) .
in other words , mass lost from the donor takes with it its contribution to @xmath50 , tending to produce contraction .
the overall @xmath50 evolution then depends on whether the underlying layers expand sufficiently to compensate for this lost radius contribution .
equation ( [ eq : drdm_pint ] ) shows that this underlying expansion depends on the two quantities : @xmath149 and @xmath150 , where @xmath151 describes the actual thermal evolution of a mass element as it is advected to lower @xmath20 . in layers where either of these quantities tends towards zero ,
the @xmath152 profile remains constant under advection , producing no net change in the relative contribution to @xmath50 from that region .
this can occur for strongly degenerate plasmas , where @xmath153 , or when the advected mass element arrives at a lower @xmath20 with the same entropy as the material it replaces .
only when the advected mass elements arrive with lower entropy ( @xmath154 ) is the relative @xmath50 contribution reduced .
if the latter case dominates in the outer layers , this can lead to a net decrease in @xmath50 .
degeneracy effects never dominate in our donors outer layers at least during this contact phase
so the @xmath50 evolution depends only on the mode of heat transport in the outer layers and on the ordering of @xmath135 and @xmath136 . when @xmath141 , heat transport has sufficient time to redistribute entropy so that @xmath155 , and little @xmath50 evolution occurs .
once @xmath156 , so that @xmath157 , the @xmath50 evolution begins to depend on the background entropy gradient . in convective regions , @xmath38 is essentially equal to @xmath158 , so that adiabatic advection leads to minimal local contributions to @xmath50 evolution . in radiative regions , however , @xmath159 ;
_ thus only for radiative surface regions during roughly adiabatic advection will a net @xmath50-contraction occur _ .
the rate of the @xmath50 contraction increases with the entropy of the layer and with the steepness of the background entropy profile .
the structure of our donor s outer layers consists of a superficial radiative layer overlying a thin outer - convective zone , followed by another radiative region that extends to the stellar centre .
the thickness of outer - convective region increases with donor degeneracy .
the entropy profile of the inner radiative region is very steep near its outer boundary ; moving inward , this entropy profile tends to flatten out rather abruptly . with this general interior structure in mind , the @xmath50 evolution trends
seen in figs .
[ fig : r_lagrange_ev ] and [ fig : r_lagrange_evb ] can be understood from the above discussion . for the @xmath124 donor ( fig .
[ fig : r_lagrange_ev ] ) , as long as @xmath141 , the @xmath50 evolution is minimal .
once this inequality is reversed , the @xmath50 evolution starts to accelerate and a rapid contraction ensues due to the steep entropy profile in the donor s radiative photosphere . by this time
, the underlying layers are advected nearly adiabatically , and the entropy gradient near the surface is continuously decreasing .
this leads to the decreasing rate of @xmath50 contraction .
finally the entropy profile becomes sufficiently shallow that the expansion of the underlying layers takes over , and @xmath50 begins increasing .
the evolution in fig .
[ fig : r_lagrange_evb ] is similar , but here the evolution is adiabatic ( @xmath140)throughout .
the @xmath50 contraction rate is slower both because the outer convective zone is thicker and the underlying radiative region has both a lower entropy and a shallower entropy profile .
this also results in less overall @xmath50 contraction .
additionally , @xmath50 decreases significantly only once a sufficient amount of mass , @xmath160 , has been lost for the pressure perturbation in the outer layers , @xmath161 , to be of order the initial @xmath20 at the top of the radiative region .
evolution for am cvn systems with @xmath162 and @xmath163 .
systems with donors having @xmath164 3.5 , 3.0 , 2.5 , 2.0 , 1.5 , and 1.1 are shown by the solid , dotted , short - dashed , dashed , short - dash dotted , and dash dotted lines , respectively .
the corresponding @xmath92 = 0.0212 , 0.0225 , 0.0248 , 0.0294 , 0.0380 , 0.0561 @xmath165 , respectively , title="fig:",width=317 ] the relative rate of @xmath50 and @xmath66 evolution characterized by @xmath166 and @xmath167determines that of @xmath15 . in all our evolution models , @xmath168 initially .
however , since @xmath169\,,\ ] ] @xmath170 rapidly decreases with growing @xmath15 .
once @xmath171 , these two quantities tend to track each other closely due to the sensitive @xmath172 dependence of @xmath15 .
thus , when the @xmath173 evolution is smooth , the @xmath15 evolution is also .
this can be seen in figure [ fig : turnon_mdot_t ] , which shows the @xmath15 time evolution for a set of models with @xmath174 , @xmath163 and differing @xmath93 ( indicated by line style with @xmath93 decreasing left - to - right ) .
the four lowest @xmath93 donors have very thin outer convective zones and their corresponding @xmath173 evolution is smooth .
the two highest @xmath93 donors ( dotted and solid curves ) have their @xmath50 response dominated , at first , by their thicker outer convective zones .
once @xmath160 begins probing the underlying radiative region , their @xmath173 suffer rapid changes in slope .
this transition from convective to radiative dominated @xmath50 response produces the non - smooth and non - monotonic behaviour seen in the dotted and solid curves .
figure [ fig : turnon_mdot_t ] illustrates other general trends .
initially , @xmath175 in all cases . while this remains true , the @xmath15 growth is approximately exponential ( corresponding to the initial , steep increase in @xmath15 ) .
lower @xmath93 donors produce slower @xmath15 growth both due to their larger contact @xmath61 ( reducing the system s @xmath176 ) and larger @xmath177 ; the latter tends to be the more significant factor . since the @xmath15 required to produce @xmath178 is greater than @xmath179 in all the systems we considered ,
all experience a phase where @xmath74 .
once this occurs , the @xmath15 growth slows as @xmath15 now depends on the donor s non - exponential , sub - photospheric @xmath18-profile ( see equation [ eq : m2dot_thick ] ) .
this results in the turn - over in @xmath15 growth seen at @xmath180-@xmath181 yr@xmath182 .
the final upturn in @xmath15 before each maximum results from a rapid increase in @xmath179 via the increasing @xmath18 of the surface layers .
this results from the adiabatic advection of lower entropy material to the surface and corresponds to the final phase of rapid @xmath50 contraction discussed above .
the extrema in @xmath50 , @xmath61 , and @xmath15 , which mark the end of the turn - on phase , do not occur simultaneously .
the @xmath50-minimum happens first and the subsequent expansion contributes to accelerating the @xmath15 growth .
the @xmath61-minimum occurs once @xmath15 reaches a critical value , @xmath183 to produce @xmath184 .
the larger @xmath185 in systems with more degenerate donors produces higher @xmath186 . eventually , the @xmath61-expansion leads to a decreasing @xmath172 , and @xmath15 reaches its maximum .
by @xmath15-maximum , the donor has lost between @xmath187 .
less degenerate donors suffer the greater mass loss .
overall the turn - on phase lasts between @xmath2 and @xmath188 yrs . by the time the system has evolved past @xmath15-maximum , the donor is evolving adiabatically in response to mass loss .
prior work on the evolution of am cvn systems in this phase has neglected the donor s thermal evolution and has either relied on predetermined @xmath13-@xmath14 tracks @xcite or has assumed the donors continue to evolve adiabatically indefinitely @xcite .
our current calculations show that at late - times , the assumption of adiabatic evolution becomes invalid ( as can be seen by the evolution of @xmath189 in fig . [
fig : r_lagrange_ev ] ) . why the donor s adiabatic evolution ends can be understood by considering how @xmath136 and @xmath135 evaluated at @xmath190 evolve under mass loss .
the quantity @xmath191 for gw driven , conservative mass transfer , the latter derivative can be written as @xmath192 where the final approximation is good when @xmath193 . in the same regime ,
@xmath194 it is useful to rewrite @xmath195 as @xmath196 where @xmath197 since both @xmath198 and @xmath4 vary only by a factor of a few during the adiabatic phase .
in contrast , @xmath49 and @xmath85 vary by roughly one and two orders of magnitude , respectively .
holding @xmath199 , @xmath4 constant , @xmath200 , and we can use hydrostatic balance to write @xmath201 where @xmath202 characterizes the @xmath85 evolution during the adiabatic phase . and
@xmath136 ( at @xmath203 ) as a function of @xmath49 for the set of am cvn models with @xmath204 and @xmath205 shown in fig [ fig : turnon_mdot_t ] .
the various lines styles have the same meaning as in that figure.,title="fig:",width=317 ] thus once @xmath193 in the adiabatic phase , @xmath206 where @xmath207 typically , @xmath208-@xmath209 and @xmath210-@xmath211 , so that @xmath212-@xmath213 and leading a rapid evolution of @xmath214 as @xmath49 is reduced . in the early portion of the adiabatic phase , where @xmath215 , @xmath216 and this scaling then gives @xmath217 when @xmath218 .
we compare this prediction to our numerical calculations in fig .
[ fig : tauthm_m ] , where we show the computed evolution of both time - scales for our models with @xmath204 , @xmath205 .
the ratio @xmath217 at @xmath219 in all cases , in excellent agreement with the scaling relation . beyond this point ,
the donor is able to shed entropy , allowing less degenerate donors to contract towards a fully degenerate configuration .
we now explore how the donor and binary evolution varies with initial conditions . in fig .
[ fig : rm_binevcomp ] , we show the @xmath220 evolution for several sets of calculations . in this figure ,
different colours correspond to different initial donor degeneracy , while different line styles indicate different @xmath221 .
the evolution along each track is from right to left , with the donors first evolving steeply downward in @xmath50 .
the amount @xmath50 decreases prior to the @xmath50-minimum depends on @xmath93 .
less degenerate donors contract to a greater extent during the turn - on phase as discussed in
[ sec : donorev_prer2min ] .
-evolution on initial binary parameters . in both panels ,
colours indicate the donor s initial @xmath222 : 1.1 ( yellow ) , 1.5 ( red ) , 2.0 ( green ) , and 3.0 ( blue ) .
line style indicates the binary s @xmath223 : 0.500 ( dashed ) , 0.825 ( solid ) , and 1.325 @xmath107 ( dot short - dashed ) .
panel ( a ) compares the @xmath50-@xmath49 evolution for systems with @xmath224 to examine the impact of initial degeneracy and donor mass .
we show evolution for systems with @xmath225 0.15 , and 0.3 @xmath107 and the initial degeneracy listed above .
panel ( b ) examines the impact of varying @xmath221 on the @xmath50 evolution .
we show comparisons at two @xmath226 to illustrate how the magnitude of @xmath221 effects are reduced at higher donor degeneracy .
, title="fig:",width=317 ] in panel ( a ) of fig .
[ fig : rm_binevcomp ] the @xmath50 evolution for a set of systems with @xmath227 are displayed . during the turn - on phase and early expansion phase ,
@xmath220 depends on both @xmath93 and @xmath89 . at fixed @xmath93 different @xmath89
have differing @xmath228 profiles , and this is reflected in the @xmath50 evolution .
the entropy differences are more significant at larger @xmath23 , producing the tendency towards convergence as @xmath49 is reduced .
near the local @xmath50-maximum ( at @xmath229 ) , track convergence is furthered by donor cooling , which erases any remaining initial @xmath32-profile information . by this point , the donors cool and contract along tracks parametrized by @xmath93 .
cooling continues and @xmath50 contraction slows as the donors become increasingly degenerate and finally reach the fully degenerate @xmath13-@xmath14 relation . in panel ( b ) of fig .
[ fig : rm_binevcomp ] , we consider the dependence of the @xmath50 evolution on @xmath221 . during the turn - on phase ,
the donor evolution is essentially independent of @xmath221 . only for the very lowest degeneracy donors
is this not the case .
these donors are able to cool somewhat during the turn - on phase and systems with lower @xmath221 produce slower @xmath15 growth , allowing greater cooling before mass loss becomes adiabatic .
the larger @xmath15 in higher @xmath221 systems also extends the duration of the adiabatic expansion phase to lower @xmath49 and larger @xmath50 . the @xmath50 path a system follows during its later contraction phase is thus parametrized by @xmath93 and @xmath221 .
-evolution on initial binary parameters .
the line styles and colours have the same meanings as in fig .
[ fig : rm_binevcomp ] . as in that figure
, panel ( a ) shows the effects of varying @xmath89 and @xmath93 at fixed @xmath221 , while panel ( b ) examines the effects of varying @xmath221 , @xmath89 at fixed @xmath93 .
here @xmath89 has values of 0.15 , and 0.3 @xmath107 .
evolution proceeds downward along each track and @xmath89 can be inferred from the initial @xmath0 within each set of constant @xmath93 tracks , with @xmath89 decreasing with increasing @xmath230.,title="fig:",width=317 ] in fig .
[ fig : tefflpo_binevcomp ] the evolution of the donor s @xmath4 is shown as a function of @xmath0 , with evolution proceeding downward along each track .
the @xmath4-evolution reflects the donor s evolutionary stages .
the significant decrease during the turn - on phase ( while @xmath0 is decreasing ) is due primarily to the removal of mass from the steep @xmath32-profile region ( see appendix [ app : l_response ] ) .
once the steep @xmath32-gradient material has been removed , the @xmath4 evolution slows along with the @xmath50 contraction . in the adiabatic expansion phase ,
@xmath4 roughly plateaus at a level set by the core entropy profile .
that is , @xmath4 during this phase is parametrized by @xmath93 and @xmath0 .
variations in the @xmath32 profiles due to @xmath89 differences are reflected in slight differences between tracks sharing @xmath93 . during this phase , @xmath23110@xmath23210@xmath233 @xmath234 , while @xmath5 also plateaus at values @xmath23510001800 k. the slow evolution in @xmath4 ends once donor cooling sets in , producing the rapid decline starting between @xmath23640 - 50 minutes . panel ( b ) shows that a larger @xmath221 extends the @xmath0 at which donor cooling start ( via an increased @xmath15 ) ; @xmath221 thus parametrizes the @xmath4 evolution during the cooling phase and beyond , with higher @xmath221 producing higher @xmath4 at fixed @xmath0 .
-@xmath0 relation on initial binary parameters .
the line styles and colours have the same meanings as in fig .
[ fig : rm_binevcomp ] . for clarity , we display the evolution of two ( @xmath88 , @xmath89 ) pairs : ( 0.350 , 0.150 ) , and ( 1.025 , 0.300 ) , where values are in @xmath107 and values of @xmath121 of 3.0 and 1.1 .
the figure focuses on the @xmath15 evolution post @xmath15-maximum .
evolution is first upward along the vertical sections of each track , then outward in @xmath0 as @xmath15 decreases .
the range of @xmath15 at fixed @xmath0 displayed here represents the range available to wd channel am cvn binaries _ given our assumptions in determining the donor s initial entropy_. , title="fig:",width=317 ] the differences in @xmath220 evolution with @xmath221 and @xmath93 are reflected in the binary s @xmath16 evolution as shown in fig .
[ fig : mdpo_binevcomp ] . for clarity , the evolution
is shown for only four systems ; these systems do , however , show the range of phase - space covered by the overall population in this study .
we show models with @xmath2373.0 and 1.1 , the latter providing a reasonable lower limit to the degeneracy expected in this population .
although not shown by this plot , the @xmath15-@xmath0 evolution during turn - on is determined primarily by @xmath89 and @xmath93 ; @xmath221 influences only the @xmath15 maximum at large @xmath93 .
once the systems have evolved into the expansion phase , the @xmath0-@xmath15 evolution depends only on @xmath221 and @xmath93 , since donors with the same @xmath93 eventually follow the same @xmath220 evolution . for gw - driven @xmath238 , at fixed @xmath0 , @xmath239 ( for @xmath240 ) so that the variations in the donors @xmath50 evolution produce a larger change in @xmath15 than do @xmath221 variations .
this trend is shown in fig .
[ fig : mdpo_binevcomp ] , where the family of curves with @xmath241 lie significantly above those with @xmath242 . within each set of curves at fixed @xmath93 ,
the spread in @xmath15 is caused by @xmath221 differences .
once donor cooling sets in , all tracks collapse toward the fully degenerate ones .
systems with lower @xmath93 and larger @xmath221 reach the fully degenerate tracks at longer @xmath0 .
after this point , the @xmath15-@xmath0 relation is parametrized exclusively by @xmath221 .
the donor s thermal evolution does not occur in isolation , but in the radiation bath provided by the flux from the accretor and disk . as the compressional heating luminosity generated in the accretor
is always much less than the accretion luminosity @xcite , the external flux seen by the donor is dominated by the accretion light .
to determine how important this external flux may be to the donor s evolution , we calculate the accretion luminosity , @xmath243 , using @xmath244 where @xmath245 , @xmath246 are the gravitational potential at the inner lagrange point and the accretor , respectively @xcite , and then compare @xmath247 to the donor s @xmath5 .
we use equations ( 14)-(16 ) of @xcite to calculate @xmath245 , @xmath246 ( noting these authors interchange our definitions of @xmath60 and @xmath49 ) . of non - irradiated donors and the @xmath248 produced by the system s accretion flow .
we show the same set of models as in figure [ fig : mdpo_binevcomp ] .
line styles and colours have the same meaning here as in that figure . for the entire range of donors ,
irradiation dominates the donor s outer boundary condition by the time @xmath15 has grown to its maximum value.,title="fig:",width=317 ] the evolution of @xmath249 versus @xmath5 for the evolution tracks from figure [ fig : mdpo_binevcomp ] is shown in figure [ fig : tirrcomp ] . during the turn - on phase , @xmath248 increases rapidly , and by @xmath15-maximum , @xmath248 has grown larger than @xmath5 by a up to a factor of 50 . in all but the lowest @xmath221 and @xmath93 cases , @xmath248 remains greater than @xmath5 after this point . in most cases , @xmath248 strongly dominates @xmath5 .
we consider now how this fact impacts the donor s and binary s evolution .
the general effect of external irradiation is to increase the temperature of the donor s atmosphere @xcite , which will tends to lower @xmath4 and slow its cooling ( see , e.g. , * ? ? ?
* ; * ? ? ?
* in the context of irradiated planets ) .
the amount of heating depends on many factors : the irradiating flux s intensity , spectrum , anisotropy , and the opacity sources in the donor s atmosphere that determine its albedo ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the detailed modelling of all these effects involves a multidimensional , non - grey , radiation - hydrodynamics problem and is beyond the scope of this paper .
thus , to model irradiation s impact , we continue to assume grey opacities and alter our temperature outer boundary matching condition to @xmath250 , where @xmath251 is defined by @xmath252 where we assume a point - source geometry for @xmath243 and complete redistribution of the irradiating flux around the donor s surface @xcite .
we also define the quantity @xmath253 , which gives the total ( intrinsic plus thermalized irradiative ) luminosity off the donor s surface .
the factor @xmath254 is a dimensionless efficiency parameter giving the fraction of @xmath243 that is thermalized in the donor s photosphere .
equation ( [ eq : irr_tsurf ] ) makes it clear that @xmath254 parametrizes ( within a grey , 1d model ) our _ entire _ ignorance associated with the above uncertainties ( including uncertainties in the assumed geometry ) . for our calculations ,
we choose a fixed @xmath255 and calculate @xmath243 from the system s _ secular _ @xmath15 .
however , am cvn binaries experience a phase in which instabilities in the he accretion disk produce cyclical variations in @xmath63 @xcite . while these outbursts produce brightness variations @xmath256 magnitudes , the outburst period is @xmath257 days @xcite .
this corresponds roughly to a @xmath135 at @xmath258 , so almost the entire donor is only aware of the time - averaged @xmath243 , making our use of the secular @xmath15 reasonable .
a star s outer boundary condition only has a significant impact on its structure when the star is nearly fully convective ( see , e.g. , * ? ? ?
further , during the adiabatic phase of am cvn evolution , mass loss , not thermal processes , dominate the donor s evolution .
thus we can expect that irradiation will affect the donor s evolution most during the cooling phase , by which time the donor s are fully convective and thermal processes dominate their evolution .
this expectation is borne out by our numerical calculations . to illustrate this , we show in figure [ fig : irr_tpcomp ] a comparison of @xmath259 profiles between an irradiated ( @xmath260 , red lines ) and non - irradiated ( @xmath261 , blue lines ) donor at several specified values of @xmath49 along respective binary evolution calculations .
apart from the @xmath254 differences , the initial conditions for the two calculations are the same .
the profiles for @xmath262 ( solid lines ) , @xmath263 ( short - dashed ) and @xmath264 ( dashed ) all occur during the turn - on and adiabatic phase .
by @xmath265 ( short - dash dotted line ) , the non - irradiated donor is in its cooling phase .
profiles for an @xmath266 irradiated ( red lines ) and non - irradiated ( blue lines ) donor at points in their evolution when @xmath262 ( solid lines ) , @xmath263 ( short - dashed ) , @xmath267 ( dashed ) , and @xmath268 ( short - dash dotted ) ; dotted portions of the lines indicate convective regions .
the initial conditions for both sets of calculations were @xmath162 , @xmath269 , and @xmath270 .
only once the non - irradiated donor is mainly convective and has begun its cooling phase ( @xmath265 ) do significant interior structural differences with the irradiated model appear . , title="fig:",width=317 ] throughout , @xmath248 dominates @xmath251 , so that the @xmath260 donor s @xmath251-evolution tracks that of @xmath15 . by increasing @xmath251 , a constant @xmath248 will tend to reduce @xmath4 @xcite . here
, there is at least one other effect important to the @xmath4 evolution .
irradiated envelopes tend to have more extensive radiative regions and steeper entropy profiles . during rapid mass loss ,
this increases the @xmath4 decrement a mass element experiences as it is advected outwards .
both effects contribute to lowering the donor s @xmath4 in the irradiated model . during phases of rapid mass loss
, the latter effect even leads to a net @xmath271 ( most obviously seen in the inverted @xmath259 profile of the @xmath272 case , but also present at @xmath273 ) .
a net @xmath271 is produced when the flux cost required to advect material up the steep entropy gradient can not be provided by donor s intrinsic flux ; the deficit is made up by absorption of irradiating flux and an inverted @xmath259 profile results . although some of the irradiating energy is absorbed below the photosphere , never more than @xmath274 of the donor is involved and the inward directed flux is always @xmath275 .
the persistence of the @xmath271 condition depends on @xmath15 and @xmath254 , with @xmath4 recovering to positive values once @xmath15 decreases sufficiently . for the case shown in fig
[ fig : irr_tpcomp ] , @xmath271 until @xmath276 , beyond the point the non - irradiated donor started its cooling phase .
figure [ fig : irr_tpcomp ] further illustrates our general results that differences between the deep interior structures of irradiated and non - irradiated donors only extend as far as the base of the convective layer , roughly speaking .
thus it is only once the donors become nearly fully convective do significant structural differences occur ; such differences become even more apparent once the non - irradiated donor begins its cooling ( e.g. , the @xmath265 profiles in fig . [ fig : irr_tpcomp ] )
. evolution for the initial conditions @xmath162 and @xmath277 .
line colour indicates @xmath121 : 1.1 ( red ) , 2.0 ( green ) , and 3.0 ( blue ) ; line style indicates @xmath254 : 0.0 ( solid ) , 0.1 ( short - dash dotted ) , and 0.5 ( dashed ) . only the non - irradiated @xmath242 case is shown due to numerical difficulties in converging the irradiated models during the turn - on phase .
however , from other irradiated models with @xmath242 the irradiated @xmath242 tracks will not differ substantially from the @xmath261 track shown . for lower @xmath93
, irradiation extends the adiabatic phase and slows the donor s cooling , elevating @xmath15 during these phases .
, title="fig:",width=317 ] in the limit of zero bond albedo for systems sharing the initial conditions of @xmath162 , @xmath163 , and @xmath164 2.0 .
line style has the same meaning as in figure [ fig : irr_mdpo ] . in the @xmath278 cases , irradiation substantially elevates @xmath279 . , title="fig:",width=317 ] by decreasing @xmath4 , the net effect of irradiation is to extend the adiabatic phase of evolution to lower @xmath49 and to slow the donor s cooling afterwards .
irradiated donors continue expanding to longer @xmath0 and contract more slowly once cooling begins .
this alters the @xmath15 evolution , as illustrated in figure [ fig : irr_mdpo ] .
the solid lines show systems evolution with non - irradiated donors , the short - dash dotted and dashed lines show @xmath280 and @xmath260 tracks , respectively .
it can be seen that differences in @xmath15 evolution between @xmath254 values appear even before the cooling phase for non - irradiated case .
this reflects the outer boundary condition s growing importance in @xmath261 donors , which have much deeper convective regions by this point in the evolution . as @xmath254 is increased , the @xmath0 where the @xmath15-decline occurs increases ; this effect is more pronounced at lower @xmath93 . before contracting to the
fully degenerate track , irradiated donors converge to an @xmath254-dependent , intermediate cooling track . in fig .
[ fig : irr_mdpo ] , this occurs at @xmath281 minutes for the @xmath282 tracks . by this point ,
the irradiated donors are fully convective , so that @xmath50 depends on @xmath283 , @xmath49 , and the strength of irradiation @xcite . as
irradiation is a function of @xmath284 through @xmath15 , the evolution of fully - convective donors within our model are parametrized exclusively by @xmath221 and @xmath254 , as evidenced by this intermediate convergence .
the evolution of @xmath279 for a set of donors with differing @xmath254 at fixed @xmath88 , @xmath89 , and @xmath93 are shown in figure [ fig : irr_teffl ] . within our grey - atmosphere modelling ,
[ fig : irr_teffl ] can be interpreted as providing the donor s surface luminosity in the limit of zero bond albedo . since @xmath248 dominates @xmath251 over most of the evolution , the @xmath279 evolution mirrors that of @xmath15 . during the turn - on phase
, @xmath279 decreases initially in all donors .
this decrease is reversed for @xmath278 by the @xmath15 growth . by @xmath15 maximum , @xmath279 @xmath285 -@xmath286 higher in irradiated models compared to non - irradiated donors . during the adiabatic phase , @xmath279 decreases with @xmath15 , while the non - irradiated donor s @xmath4 plateaus . in the cooling phase @xmath279 converges towards tracks parametrized by @xmath254 , but along much shallower slopes than non - irradiated donors
. even so , by @xmath287 minutes , even @xmath260 donors are still rather dim , with @xmath288 .
we start our discussion by comparing the @xmath16 evolution produced in our current donor models to that produced by several prior donor models in figure [ fig : mdpo_priormod_comp ] .
the lower panel focuses on the comparison between our current models ( solid lines ) and the @xcite isentropic models ( dashed lines ) .
the @xcite models assume the donors are fully convective and application of these models in both @xcite and @xcite also assume adiabatic donor evolution .
our more complete , current modelling shows that these he donors are not fully convective over much of the am cvn evolution phase and that adiabatic evolution only occurs out to @xmath289 minutes .
the impact of these differences is apparent between the @xmath15 evolution shown by the dashed and solid lines in fig .
[ fig : mdpo_priormod_comp ] . the initial conditions ( i.e. @xmath290
, @xmath49 , @xmath50 ) for each dashed line evolution equals the set of these values along the corresponding solid line at the point of intersection . from there
, the isentropic donors produce @xmath15 evolution that increasingly diverges upwards from the solid - line tracks , even during the adiabatic evolution phase .
this results from the realistic models being radiative throughout much of their core .
thus , compared to isentropic donors , the realistic models have a greater @xmath291 , even during adiabatic evolution ( [ sec : donorev_prer2min ] and appendix [ app : r_response ] ) .
the differences between these evolution tracks become even more dramatic once the realistic donors begin cooling . thus , modelling ultracompact binary evolution using isentropic donors will overestimate the @xmath50 expansion rate , the @xmath15-@xmath0 relations , and @xmath292 .
-@xmath0 evolution produced by different donor models .
the lower panel compares evolution with our current models ( solid lines ) to that with the @xcite isentropic models ( dashed lines ) .
the two solid lines had @xmath293 and @xmath294 ( upper ) and @xmath295 ( lower ) ; in all four systems @xmath296 .
the initial conditions ( @xmath60 , @xmath49 , @xmath50 ) for each dashed line evolution equalled the set of these values on the corresponding solid line at the point of intersection .
the upper panel compares evolution with our current models ( solid lines , lower same as in lower panel , upper with @xmath297 , @xmath298 , @xmath241 ) with the evolution produced by the semi - degenerate donor @xmath13-@xmath14 relation of @xcite with @xmath299 and @xmath300 ( red line ) .
[ fig : mdpo_priormod_comp],title="fig:",width=317 ] this has implications for how @xmath15 measurements constrain an am cvn system s formation channel .
@xcite showed that assuming isentropic donors leads to a significant overlap in the @xmath15-@xmath0 plane between hot wd channel systems and he - star channel systems .
the overlap between these channels is much reduced by with our current donor models , as illustrated in the upper panel of fig .
[ fig : mdpo_priormod_comp ] .
there the black lines show the maximum range in the @xmath15-@xmath0 relation for wd channel am cvn systems produced with our current unirradiated models .
the red line shows the @xcite fit to a semi - degenerate donor @xmath301 evolution with @xmath302 , and provides a rough lower limit to he - star channel systems @xmath16 evolution .
only the most massive wd channel systems have any overlap with he - star channel systems .
thus , given this current theory , a determination of a secular @xmath15 significantly above the upper black fig .
[ fig : mdpo_priormod_comp ] would indicate a system formed through the he - star channel .
this statement has one caveat : if wd channel systems have a maximum entropy greater than indicated by our determinations in [ sec : initial_models ] or if the donors are heated significantly ( e.g. , by tidal mechanisms ) earlier in the am cvn phase , then the overlap with the he - star channel systems could be increased .
we point out that modelling he - star channel systems with @xcite semi - degenerate @xmath301 relation beyond @xmath303 is problematic .
this is because these donors will also begin cooling and contracting by these @xmath0 , similar to the wd channel donors .
thus , in reality , the red line in fig [ fig : mdpo_priormod_comp ] should begin a down turn somewhere in the vicinity of the @xmath304 track .
in fact , the original calculation used by @xcite to determine this fit ( model 1.1 of * ? ? ?
* see their fig .
2 ) shows the start of this down turn at @xmath305 minutes .
thus , beyond the @xmath0 at which donors begin their cooling , @xmath15 measurements alone will not distinguish between formation channels .
while the changes to the adiabatic phase @xmath284 evolution will quantitatively alter our expectations for the am cvn population s @xmath0-distribution , the detailed evolution to @xmath0-minimum and the occurrence of donor cooling will produce qualitatively new features in the population s @xmath0-distribution .
we discuss these features here .
a key factor in how an individual system contributes to the @xmath0-distribution is the time - derivative , @xmath306 , of its @xmath0-evolution . in a steady - state
, continuity requires that the number density of systems at some @xmath0 , @xmath307 , scale as @xmath308 ( see , e.g. , * ? ? ?
here we will use this scaling relation to display steady - state distributions of systems sharing initial data .
this provides a straightforward means for displaying how differing initial conditions influence the relative contribution systems make to the overall @xmath0-distribution .
we calculate @xmath306 along an evolutionary track via @xmath309\ , .
\label{eq : pdot}\ ] ] we show the evolution of @xmath306 about the @xmath0-minimum for a representative set of systems in figure [ fig : pdotpcontact ] .
these systems have @xmath310 , @xmath311 , with colours indicating different @xmath93 .
the short - dash dotted segments indicate @xmath312 , solid segments @xmath313 .
the black line shows the inward evolution due only to the @xmath314 term in equation ( [ eq : pdot ] ) for comparison . as discussed in
[ sec : donorev_prer2min ] , the evolution here is most sensitive to @xmath89 and @xmath93 .
the @xmath306 evolution reflects this , with @xmath93 ( as illustrated in this figure ) and @xmath89 both affecting the the value of the @xmath0 minimum and how @xmath306 diverges from the gw - only evolution once contact occurs .
the generic features of our this @xmath306 evolution are the strong spikes at @xmath0-minimum as @xmath306 evolves through zero and the existence of @xmath306 minimum and maximum that occur before and after @xmath0-minimum , respectively . about the @xmath0-minimum for representative systems with @xmath315 and @xmath316 .
line colour indicates @xmath121 : 1.1 ( yellow ) , 1.5 ( red ) , and 3.0 ( blue ) .
short - dash dotted line segments indicate @xmath317 , solid lines indicate @xmath318 .
the black line shows the corresponding inward @xmath306 evolution produced by gw emission alone.,title="fig:",width=317 ] the relative magnitude of @xmath306 before and after the @xmath0-minimum which determines the relative number of systems evolving inward vs. outward in a steady - state depends on @xmath89 and @xmath93 .
the @xmath312 just before @xmath306-minimum is approximated by the gw - only rate , which scales as @xmath319 .
by @xmath306-maximum , @xmath15 has achieved its secular rate : @xmath320 producing a @xmath306 during expansion : @xmath321 \equiv 3 \left({\frac{\dot{j}}{j}}\right)_{\mathrm{gw } } \beta_{\dot{p}}\ , .
\label{eq : pdot_eq}\ ] ] if @xmath49 does not change appreciably from @xmath306-minimum to @xmath306-maximum , the term in square brackets , denoted here by @xmath322 , estimates the relative magnitudes of @xmath306 before and after @xmath0-minimum . note that @xmath322 depends on @xmath93 through @xmath173 . during the early expansion phase
, @xmath173 evolves from @xmath323 to @xmath324 to @xmath325 . with @xmath326 ( the @xmath327 range considered here ) ,
@xmath322 can vary considerably .
typically , @xmath328 , as seen in fig .
[ fig : pdotpcontact ] .
however , for larger @xmath327 and @xmath93 , @xmath322 can be much less than -1 ; we see this behaviour in our calculations with @xmath329 , @xmath293 for @xmath330 .
lower @xmath93 donors have smaller relative outward - to - inward @xmath331 , for two reasons .
one , lower @xmath93 leads to larger minimum @xmath173 , producing slower expansion and @xmath322 values closer to zero .
two , @xmath49 during the @xmath306-transition is not fixed , with @xmath49 losses greater for lower @xmath93 .
this results in a lower @xmath238 loss rate post @xmath306-maximum , further contributing to smaller @xmath332 . and @xmath333 with colour indicating @xmath121 : 1.1 ( yellow ) , 1.5 ( red ) , 2.0 ( green ) , 3.0 ( blue ) , and 3.5 ( cyan ) .
the lower panel separates systems by @xmath306 sign : the short - dash dotted histograms counts inward moving systems , the solid histograms outward moving systems .
the upper panel displays the sum of both inward and outward moving systems . the black short - dash dotted line shows the distribution for the evolution produced by gw emission alone ( i.e. @xmath334 ) .
the bins have width of @xmath335 s. the overall normalization is arbitrary , but the relative normalization accurately reflects the relative @xmath306 rates . , title="fig:",width=317 ] how the evolution near @xmath0-minimum translates into the steady - state @xmath0-distributions is shown in figure [ fig : nhist_400 ] .
there we display histograms of relative system numbers along evolutionary tracks with @xmath310 , @xmath336 .
the number in each bin , @xmath109 , is calculated by integrating @xmath337 over each bin .
the overall normalization is arbitrary but fixed across tracks , so that the histograms accurately reflect relative @xmath306 rates .
the lower panel displays separately the contributions from the @xmath338 ( short - dash dotted lines ) and @xmath313 ( solid lines ) segments of the evolution .
the black line in the lower - panel shows the corresponding result for a gw - only driven in - spiral .
the upper panel shows the sum of contributions from both inward and outward evolving systems .
the general trends in the lower panel are the slight deviation from gw - only evolution from contact inward , followed by a peak as @xmath339 at @xmath0-minimum .
the outwardly evolving systems likewise show peaks just post @xmath0-minimum . for larger @xmath0 ,
lower-@xmath93 tracks lead to fewer systems at a given @xmath0 since hotter donors have larger @xmath49 , producing larger @xmath319 ( * ? ? ?
* see also fig .
[ fig : pdotpcontact ] ) .
the sharp steps in the upper - panel histograms result from starting our calculations at the point of contact ( i.e. pre - contact evolution is not included ) , so the location and size of each step is somewhat artificial .
however , since they result from having a definite starting point for the gw - driven in - spiral , as is provided by the ce - event forming these systems , there is a physical basis for expecting their existence .
the most striking feature of these histograms are the strong peaks at each system s @xmath0-minimum .
the impact of these features on the integrated am cvn @xmath0-distribution below @xmath340 minutes will depend on the distribution of initial conditions , the survival of systems at contact ( see , e.g. , * ? ? ?
* ) , and how he - star channel systems contribute in this @xmath0 range .
determining how the integrated @xmath0-distribution depends on initial conditions and the physics determining the outcomes at contact is the subject of current work .
this will be most relevant to future space - based gw - interferometers , such as _ lisa _ , which will provide a rather complete census of the galactic am cvn population at these @xmath0 @xcite and offer direct observational tests of these predictions . in @xcite
, we considered the relevance of the @xmath312 phase to the short period x - ray variables , rx j0806 + 1527 and rx j1914 + 2456 ( 569 s ; * ? ? ?
there is still some question as to the nature of these two sources .
if both system s periods are orbital and their measured @xmath306 @xcite secular , then @xcite showed that both are consistent with being members of the am cvn population .
if this is the case , there is the question of whether observing either system in its present state is an extremely unlikely event .
a detailed answer to this question is beyond the scope of this paper .
however , a related question can be posed : what is the relative likelihood of detecting each system in its @xmath312 versus @xmath332 phase ?
figure [ fig : rxsys_pdotp ] provides a partial answer .
the lower panel shows the @xmath306 evolution for two sets of am cvn systems , each of which bracket the evolution of systems consistent with either rx
j0806 + 1527 or rx j1914 + 2456 ( see * ? ? ?
the upper panel shows , for these evolution tracks , the ratio of the number of systems with @xmath332 to those with @xmath338 in steady - state .
this ratio is only @xmath341 for periods below each system s measure @xmath0 .
thus , there is not a strong a priori bias to detecting such systems in their @xmath332 phase , although the actual relative detection likelihood between the two phases is likely influenced strongly by selection effects .
minutes ) and rx j1914 + 2456 ( lines at @xmath342 minutes ) .
the set of two lines shown for each system indicates the approximate range of variation allowed by observations and our modelling .
different colours are meant only to guide the eye in matching tracks within and between the two panels .
the lower panel shows the @xmath306 evolution for the selected tracks , with solid lines indicating @xmath313 and short - dash dotted lines @xmath312 .
the upper panel shows the ratio of of number of systems with @xmath332 to those with @xmath338 assuming steady - state .
the dotted lines indicates the measured @xmath0 of each system , while the diamonds show their measured @xmath306 . in making these comparisons , we explicitly assume the measured @xmath306 reflects the system s _ secular _ @xmath306 .
, title="fig:",width=317 ] we now turn to how donor cooling influences the @xmath0-distribution . by the time @xmath306 has passed its maximum , it is evolving at a rate given by equation ( [ eq : pdot_eq ] ) , and is essentially determined by @xmath15 ( which is a function of @xmath221 and @xmath93 ) .
this is seen in the lower panel of figure [ fig : pdotp_nhist_outward ] where we display @xmath306 versus @xmath0 for two different @xmath221 at two different @xmath93 .
before systems begin their cooling phase , the ordering of tracks is for lower @xmath93 donors to produce higher @xmath306 .
after the cooling phase ends , tracks are distinguished only by @xmath221 . during cooling
, the donor s contraction stalls the @xmath0 evolution , producing a reversal in the ordering of @xmath306 with @xmath93 and a distinctive peak in the steady - state @xmath0 distribution along each track ( upper panel of fig .
[ fig : pdotp_nhist_outward ] ) . for @xmath343 min during the expansion phase .
the lower panel shows @xmath306 vs. @xmath0 for two sets of @xmath221 each with two different @xmath121 .
colours indicate @xmath121 and have the same meaning as in fig .
[ fig : nhist_400 ] .
dashed lines show systems with @xmath344 , @xmath345 ; short - dash dotted lines systems with @xmath346 , @xmath347 .
the upper panel shows the relative number of systems expected in steady - state along each evolution track . as in fig .
[ fig : nhist_400 ] , the overall normalization is arbitrary , but the relative normalization between tracks is set by their @xmath306 rates .
the slowing of @xmath306 during the donor s cooling phase leads to a peak in system numbers .
the peak s @xmath0 is diagnostic of @xmath221 ( as well as @xmath254 , although this is not show here ) .
the magnitude of the peak increases with decreasing @xmath93.,title="fig:",width=317 ] the location and magnitude of these peaks are determined by @xmath221 and @xmath93 .
the system s @xmath221 determines the @xmath0 at which cooling becomes important , with larger @xmath221 moving the peak s centre to longer @xmath0 .
the donor s @xmath93 determines the degree of donor contraction ; hotter donors lead to slower @xmath306 during the cooling phase and larger peaks . in slowing donor cooling
, irradiation also acts to shift the peak centres to longer @xmath0 .
the distribution of am cvns above @xmath305 minutes will provide an integrated diagnostic of the distribution of @xmath93 , @xmath221 , and @xmath254 in this @xmath0 range , am cvn binaries will not be individually resolvable gw - sources due to the galactic foreground of detached wd - wd binaries @xcite .
thus , whether the system distribution in this @xmath0 range can be a practical diagnostic tool must await observations progress in the optical / ir wave bands .
although we reserve detailed applications of the models presented here to specific am cvn systems for a later companion paper , we shall discuss the recently discovered eclipsing am cvn binary , sdss j0926 + 3624 @xcite as a system for which the entropy of the donor can be probed .
this is especially important as the distribution of donor entropy provides ( i ) a potential diagnostic of am cvn formation channels and ( ii ) constrains the stellar binary evolution determining am cvn initial conditions @xcite .
@xcite have presented evidence based on accretor properties that the two known long - period am cvn binaries , gp com ( @xmath348 minutes ) and ce-315 ( @xmath349 minutes ) both harbour a relatively hot donor . additionally ,
if early - contact am cvn binaries are indeed the correct model for the systems rx j0806 + 1527 and rx j1914 + 2456 , then this would provide further evidence for hot donors : the bracketing models shown in fig . [
fig : rxsys_pdotp ] , have a range of @xmath350 for rx j0806 + 1527 and @xmath351 for rx j1914 + 2456 .
sdss j0926 + 3624 , however , has finally provided direct evidence for hot donors in am cvn binaries .
this evidence comes in the form of @xmath60 and @xmath49 determinations made via modelling of eclipse light - curves @xcite . assuming the accretor obeys a fully degenerate wd @xmath13-@xmath14 relation ( a good approximation at the determined @xmath60 ) , these authors determine @xmath352 and @xmath353 , giving @xmath354 , @xmath355% more massive than a zero - temperature wd that fills its roche lobe at this system s @xmath356 minutes .
this @xmath49 measurement provides us the first direct means of determining an am cvn donor s current entropy . from the determined @xmath49-range , we find that @xmath121 lies approximately in the range 1.601.35 .
if the actual range of he - star channel donor @xmath13-@xmath14 evolution does not differ significantly from the @xcite fit for these systems , then a he - star channel system should have @xmath357 at @xmath358 minutes .
the evolve - ms channel also appears to produce a value of @xmath49 which is too high for systems at this @xmath0 ( see table 1 of * ? ? ?
_ thus sdss j0926 + 3624 presents evidence that the wd channel indeed contributes to the observed am cvn population and that this channel produces non - zero entropy donors as predicted by @xcite_. we can also predict the current @xmath15 in sdss j0926 + 3624 . from our calculations ,
a system with the determined value of @xmath359 and @xmath360 at @xmath358 minutes , has an @xmath361 yr@xmath182 . at fixed @xmath0 when @xmath193 , @xmath362 @xcite , so the error bars quoted in @xcite for @xmath60 and @xmath327 provide an error of @xmath363 on @xmath15 .
thus , we estimate @xmath364 yr@xmath182 .
this @xmath15 range is close to value at which the accretor s thermal evolution decouples from the compressional heating provided by the accretion @xcite .
thus , we can use fig . 1 from @xcite to estimate the accretor s @xmath85 from @xmath15 and find @xmath365-@xmath366 k. using these @xmath15 and @xmath85 ranges
, we sum the accretor s cooling and compressional luminosity ( see 2 of * ? ? ?
* ) to estimate the accretor s @xmath367-@xmath368 k , taking @xmath369 to determine @xmath370 .
more refined estimates of the accretor s thermal properties will require more detailed calculations taking into account variations in @xmath15 evolution histories , the time - dependent evolution of the accretor s envelope during the decoupling phase , and possibly the effect of he shell flashes on the accretor s surface .
the measurements of @xmath221 and @xmath93 in additional systems are required before the observational distributions of these parameters can be determined . however , the likelihood of discovering a system such as sdss j0926 + 3624 given our theoretical models of the wd channel am cvn population can be considered .
we first compare @xmath221 to the distribution in fig . 1 of @xcite .
the locus of points defining the region @xmath371 in this figure lies outside the @xmath60-@xmath49 parameter space these authors consider most likely .
this raises the question of whether observing such a high @xmath221 in sdss j0926 - 3624 simply is the result of small number statistics or whether this is a hint of additional physics that skews the @xmath221 distribution to higher values .
an example of such physics would be the preferential survival of high @xmath88 systems during a direct impact accretion phase at contact @xcite .
finally , from figure [ fig : psici_hist ] , about 12% of wd channel am cvn systems in our modelling have @xmath93 in the range consistent with sdss j0926 + 3624 , not taking into account the system s high @xmath221 . given the rather flat @xmath93 distribution expected from theory , determining @xmath221 and @xmath91 constraints in additional am cvn systems is essential if we are to determine if this is consistent with our theoretical expectations or not .
here we consider predictions for the donor s contribution to the system s light in sdss j0926 + 3624 based on our current models .
the near edge - on inclination of sdss j0926 + 3624 makes this system a good candidate for discriminating the accretor s and donor s light without significant contamination from the accretion disk .
the discovery spectrum of sdss j0926 + 3624 @xcite is reproduced in figure [ fig : j0926_spectra ] from the sdss archival data .
it was obtained over a 3600s exposure roughly two system orbits and thus provides a phase averaged spectrum .
it is decidedly non - blackbody in shape , indicating that a db wd accretor is not the only contributor to the system s light .
k , @xmath372 accretor ( red line ) and @xmath373 k , @xmath374 donor ( which lies below the x - axis ) at a distance of 293 pc .
the lower panel shows the combined spectrum ( yellow line ) from a @xmath375 k accretor ( red line ) and a steady - state @xmath376-disk model ( blue line ) with @xmath377 yr@xmath182 at a distance of 695 pc.,title="fig:",width=317 ] the question is whether the second component contributing to the flux is the donor or the disk ?
the @xcite @xmath60 , @xmath49 constraints imply @xmath378 and @xmath379 . from our current modelling
, we predict the donor s @xmath380-@xmath381 k , corresponding to an @xmath254 range of @xmath382-@xmath383 . with @xmath384 and @xmath385 @xcite , the projected area of the disk s face @xmath386 ; i.e. , out of eclipse roughly equivalent areas of disk and
donor surfaces are seen .
thus , the disk may provide a significant contribution to the system s light .
indeed , @xcite find that the disk contributes @xmath355% of the flux in the @xmath387 band and @xmath388% in the @xmath389 .
they also find significant inter - orbit variability , so whether this particular ratio of flux contributions is representative of conditions during the sloan observation is not certain . given the prominent double - peaked he emission lines in this system , am cvn phenomenology
would argue the disk is either in a stable , seemingly optically thin , low - state or is an outbursting disk caught in quiescence . in either case , a stable , optically thick @xmath376-disk model spectra is not expected . to check this , we calculated an @xmath376-disk spectra assuming @xmath390 yr@xmath182 and an outer radius of @xmath391 .
we added this to the accretor s flux modelled as a single @xmath5 blackbody .
we then adjusted this @xmath5 and the system s distance to give a `` by eye '' best fit to the sdss j0926 + 3624 spectrum .
the results are shown in the lower panel of figure [ fig : j0926_spectra ]
. a rather hot , @xmath375 k , accretor and a system distance of 695 pc is required for a reasonable fit .
this model underestimates the continuum flux at wavelengths @xmath392 and @xmath393 .
additionally , the @xmath387 flux from the disk is significantly greater than the accretor , in disagreement with @xcite .
this however may not be a significant issue given the source s variability .
we also considered a model with accretor and donor contributions , but no disk .
both components were modelled as single @xmath5 blackbodies . the observed spectra is well fit by the combination of a @xmath394 k accretor and a @xmath395 k donor at a distance of 293 pc ( upper panel of figure [ fig : j0926_spectra ] ) .
the accretor s @xmath5 is a little below the range predicted by the @xcite theory ( [ sec : discussion_hot_donors ] ) , while the donor s @xmath5 is consistent with an @xmath254 somewhat below 0.5 .
the fit in this case is somewhat better than the accretor+disk model .
we should note that other acceptable fits can be found by collectively increasing or decreasing both components temperatures and the system s distance .
thus , higher accretor @xmath5 require higher @xmath254 values for the donor to be hot enough to produce a good fit to the spectra .
a parallax measurement for sdss j0926 + 3624 will clearly distinguish between the stable @xmath376-disk+accretor and the accretor+donor model ( as well as further constrain the component temperatures in this latter model ) . the better agreement between data and donor+accretor model and our expectation that the disk is not in a stable high - state already argues for the donor+accretor model .
given our poor understanding of a low - state disk spectra and flux , however , one can only conclude that an accretor+donor model is fully consistent with both the observed spectrum and all theoretical expectations for sdss j0926 + 3624 .
since this model does nt include any disk contributions , it does fail to explain the inferred disk properties of the @xcite results .
it may be possible that it is a cool disk providing the long-@xmath396 flux in this system , possibly explaining both the sloan spectrum and the @xcite results .
better understanding of quiescent he disks and phase - resolved spectral studies of this system may both be required to break this degeneracy between either a cold disk or donor as the source of long wavelength flux .
we have implemented a new stellar evolution code in c++ that allows significant flexibility in use of input physics and defining systems of odes .
this code has been used to model , for the first time , the full stellar structure of the donors in wd - channel am cvn binaries .
specifically , the thermal and structural evolution of the donor has been calculated to determine how the donor s thermal evolution affects the evolution of the binary and to provide the first predictions for the donor s light in these systems .
systems forming through the wd channel are expected to have a range of donor properties , most importantly a range of initial ( i.e. , at contact ) degeneracy @xcite .
we modelled the pre - contact evolution of wd channel systems based on the @xcite population synthesis to determine suitable initial donor models for our subsequent calculations .
the donors in this population have initial central degeneracy parameters between @xmath39710-@xmath81 , with the distribution in @xmath93 being rather flat between @xmath39725 - 4000 .
most of these systems make contact at @xmath2362 - 11 minutes .
this range of donor parameters is likely dependent on assumptions about the binary evolution leading to their formation ( in particular the ce - event prior to the am cvn phase ) and our assumptions about the donor s cooling during the pre - contact phase .
quantifying the potential effects of these assumptions will require more extensive modelling of the proto - donors in this pre - contact phase .
our evolutionary calculations show that wd - channel am cvn systems have three phases of evolution .
an @xmath15 turn - on phase , during which @xmath15 grows to its maximum value while @xmath50 contracts under mass loss , and @xmath0 decreases .
this behaviour produces a turn - on phase lasting significantly longer than previous estimates ( e.g. , * ? ? ?
* ; * ? ? ?
* ) : @xmath2-@xmath3 yrs depending on the donor s initial @xmath93 . in the second phase , the donor expands adiabatically under mass loss .
the third phase begins once the mass loss rate and the donor s thermal time have have both decreased enough to allow , starting at @xmath398 minutes , the donor to cool and contract to a fully degenerate configuration .
we discussed how the system s initial conditions influence the later evolution of @xmath220 and @xmath16 .
we also revised the upper limit to the @xmath399 relation for wd - channel systems ( given our initial condition determinations ) .
finally , we predicted the donor s intrinsic @xmath4 and @xmath5 ; during the adiabatic phase , @xmath23110@xmath23210@xmath233 @xmath234 , while @xmath40010001800 k. the flux generated by the accretion flow in these systems can easily dominate the donor s intrinsic thermal output .
we self - consistently modelled the impact of the accretion light on the donor s and binary s evolution in the grey approximation .
the irradiation reduces the donor s intrinsic @xmath4 , producing a delay in the onset of cooling and slowing the donor s contraction once cooling does begin .
this shifts the downturn in the @xmath16 relation seen in the non - irradiated donors to longer @xmath0 .
irradiation also elevates the donor s photospheric temperature and surface luminosity ( up to 10,000 k and @xmath401 , respectively , during the adiabatic phase ) .
the observational signatures of irradiation depend on the irradiating flux s spectra , the opacity sources in the donors atmosphere and the efficiency of day - to - night side energy flow , all considerations beyond this work s scope .
our predictions for the irradiated donor s light are thus approximate and valid in the limit of complete redistribution of flux in a grey atmosphere with a bond albedo of zero .
in comparison to prior predictions using simpler donor models , we find that previous assumptions of fully - convective donors and adiabatic evolution @xcite are not valid over much of the am cvn evolution .
the donors only become fully convective at @xmath305 minutes and @xmath402 . prior to this
, their shallower entropy profile leads to a slower expansion rate relative to fully convective models . due to this
, we argue that prior to the onset of cooling , @xmath15 measurements could distinguish between systems formed in the he - star versus wd channels ( as opposed to the conclusions of * ? ? ?
. after the cooling phase develops , all initial donor entropy information , including any that distinguishes formation channels , will be erased .
a system s evolution about its @xmath0-minimum and the occurrence of donor cooling phase both leave diagnostic signatures on the am cvn population s @xmath0-distribution .
most significantly , the evolution of @xmath339 at the @xmath0-minimum and the slowing of @xmath0 evolution that occurs during the cooling phase lead to peaks in the @xmath0-distribution whose location and size depend on initial parameters ( and @xmath254 in the case of the cooling peaks ) .
these could provide observational diagnostics of the distribution of these parameters in the galactic am cvn population .
finally , we showed that recent measurements of @xmath60 and @xmath49 in the eclipsing am cvn system , sdss j0926 + 3624 @xcite provide direct evidence that wd - channel systems contribute to the am cvn population and that this channel produces non - zero entropy donors as predicted by @xcite .
we compared predictions for this system s light based on our models , showing that a composite spectrum consisting of donor and accretor contributions is fully consistent with this system s discovery spectra .
based on this , we predict this system lies at a distance of @xmath403 pc .
current investigations include applying this theory to interpret the improving observational constraints in many of the known am cvn systems ( e.g. , roelofs et al .
2007 , submitted ) .
in addition , we are considering how the physics relevant to the am cvn system formation and early contact - phase survival could be probed by future _ lisa _ observations of the @xmath0-distribution of sources at @xmath404 minutes .
other indicated work includes proper non - grey , phase dependent modelling of the donor s irradiated atmosphere as well as progress on understanding the contribution of low - state he accretion disks to the system s emission .
we thank the anonymous referee for a careful reading of this manuscript and for suggestions that improved its overall presentation .
we thank susana barros , gijs roelofs , tom marsh , gijs nelemans , bart willems , and dean townsley for helpful discussions and encouragement during the preparation of this manuscript , as well as jason alexander for providing the low - temperature opacity tables needed in this work and for answering associated questions . for material support ,
cjd thanks craig heinke , vicky kalogera , and greg and lauree hickok , without whose help this work could not have been completed .
this work was supported by nasa through xmm grant nnx06ah62 g and by the nsf through grants ast-0200876 and ast-0449558 .
the donor s @xmath50 can be expressed as the integral of equation ( [ eq : drdm ] ) : @xmath405 the derivative of equation ( [ eq : rdmint ] ) with respect to @xmath49 is @xmath406\\ & = \frac{1}{4 \pi r_2 ^ 2 } \left [ \frac{1}{{\rho_{\mathrm{phot } } } } - \int_0^{m_2 } \frac{dm}{\rho } \left(\frac{1-\chi_t \nabla'}{\chi_{\rho } } \right ) \left({\ensuremath{\frac{d \,\ln p}{dm_2}}}\right)_m \right]\ , , \label{eq : drdmint } \end{split}\ ] ] where @xmath407 , @xmath408 and @xmath151 gives the change in @xmath19 at fixed @xmath23 that occurs due to changes in @xmath49 .
the quantity @xmath409 , showing the donor s outer layers dominate the @xmath50 response to @xmath49 . also , @xmath410 and @xmath411 generically in our donors , so the local response to mass loss at fixed @xmath23 is an expansion in @xmath21 .
the surface term then is the only driver of @xmath14 decrease under mass loss ( see also * ? ? ?
note that both these behaviours are apparent in fig .
[ fig : r_lagrange_ev ] . how the @xmath50 evolution depends on the donor s structure is more easily seen by transforming to @xmath20-coordinates and considering only the donor s outer layers .
the contribution , @xmath412 , to @xmath50 between surface at @xmath413 and some pressure @xmath414 : @xmath415 where @xmath416 const .
is the gravitational acceleration at @xmath417 .
the change in this layer s thickness under mass loss is then : @xmath418 where we have neglected for simplicity the surface term s contribution . the layer s response the depends on @xmath419 , which can be rewritten as @xmath420 \left ( { \ensuremath{\frac{d \,\ln p}{dm_2}}}\right)_m\\ & = \frac{\chi_t}{\chi_{\rho } } \left ( \nabla - \nabla ' \right ) \left ( { \ensuremath{\frac{d \,\ln p}{dm_2}}}\right)_m\ , .
\label{eq : drdm_pint } \end{split}\ ] ]
to examine how the donor s @xmath22-profile evolution in response to mass loss depends on the donor s structure and relative ordering of @xmath135 and @xmath136 , taking @xmath421 we can rewrite equation ( [ eq : dldm ] ) as @xmath422 where the first approximation holds when mass loss effects dominate @xmath423 and the elapsed time under consideration @xmath424 . in the outer layers , the region most heavily weighted by the @xmath425 term ( see appendix [ app : r_response ] ) ,
this last expression can be expressed approximately as @xmath426 where @xmath427 , heat transport is able to approximately maintain the original thermal profile , producing @xmath155 . for radiative regions , @xmath159 ,
producing @xmath428 as mass elements must absorb flux in order to increase their entropy as they move outward . in convective regions , @xmath429 .
since the fractional change in @xmath22 , @xmath430 due to advection goes as @xmath431 , the overall flux decrement is rather small in this limit . in the opposite limit , when @xmath140 , mass elements are advected outward nearly adiabatically .
thus @xmath157 from below in radiative regions and from above in convective zones .
since @xmath214 can become very large , significant @xmath22 perturbations due to mass loss can be driven by only a very slight non - adiabaticity in the advective flow .
our numerical calculations bear this out , showing that the magnitude of @xmath22 perturbations are largest during the adiabatic mass loss phases .
|
we calculate the full stellar - structural evolution of donors in am cvn systems formed through the wd channel coupled to the binary s evolution . contrary to assumptions made in prior modelling , these donors are not fully convective over much of the am cvn phase and do not evolve adiabatically under mass loss indefinitely . instead , we identify three distinct phases of evolution : a mass transfer turn - on phase ( during which @xmath0 continues to decrease even after contact , the donor contracts , and the mass transfer rate accelerates to its maximum ) , a phase in which the donor expands adiabatically in response to mass loss , and a cooling phase beginning at @xmath1 4555 minutes during which the donor contracts .
the physics that determines the behaviour in the first and third phases , both of which are new outcomes of this study , are discussed in some detail .
we find the overall duration of the turn - on phase to be between @xmath2-@xmath3 yrs , significantly longer than prior estimates .
we predict the donor s luminosity , @xmath4 , and effective temperature , @xmath5 . during the adiabatic expansion phase ( ignoring irradiation effects ) , @xmath6@xmath7 and @xmath8@xmath9 k. however
, the flux generated in the accretion flow dominates the donor s intrinsic light at all times .
the impact of irradiation on the donor extends the phase of adiabatic expansion to longer @xmath0 , slows the contraction during the cooling phase , and alters the donor s observational characteristics .
irradiated donors during the adiabatic phase can attain surface luminosities up to @xmath10 .
we argue that the turn - on and cooling phases both will leave significant imprints on the am cvn population s @xmath0-distribution .
finally , we show that the eclipsing am cvn system sdss j0926 + 3624 provides evidence that wd - channel systems with non - zero entropy donors contribute to the am cvn population , and we discuss the observational signature of the donor in this system .
binaries : close gravitational waves stars : individual ( rx j0806 + 1527 , rx j1914 + 2456 , sdss j0926 + 3624)white dwarfs
|
differential forms and sheaves of differentials are fundamental objects and indispensable tools in the study of smooth varieties and complex manifolds .
it is well - known that for singular spaces , there is no single notion of `` differential form '' that captures all features of the smooth case .
instead , there are several competing definitions , each generalising certain aspects .
the following two classes of differential forms are particularly important .
khler differentials : : uniquely determined by universal properties , khler differentials are the most fundamental notion of differential form .
their universal properties imply that khler differentials can be pulled back via arbitrary morphisms , and closely relate them to problems in deformation theory .
+ the sheaf of khler differentials on a scheme @xmath3 is denoted @xmath7 . as a sheaf of @xmath8-modules
, @xmath7 is usually rather complicated and has both torsion and cotorsion , even if @xmath3 is reduced .
this makes khler differentials rather hard to work with in many cases of practical interest .
reflexive differentials : : given a normal complex variety @xmath3 , a _
reflexive differential _ on @xmath3 is a differential form defined only on the smooth locus , without imposing any boundary condition near the singularities .
equivalently , a reflexive differential is a section in the double dual of the sheaf of khler differentials . denoting the sheaf of reflexive differentials by @xmath9}_x$ ]
, we have @xmath10}_x = \bigl ( ^p_x \bigr)^ { * * } = _ * \bigl ( ^p_{x_{\operatorname{reg } } } \bigr),\ ] ] where @xmath11 denotes the inclusion of the smooth locus .
since the @xmath8-module structure is comparatively simple , reflexive differentials are often quite useful in practise .
reflexive differentials arise naturally in a number of contexts , for instance in positivity results for differentials on moduli spaces .
note that the dualising sheaf is always a sheaf of reflexive differentials , @xmath12}_x$ ] .
reflexive differential forms do not generally satisfy the same universal properties as khler differentials .
however , it has been shown in a previous paper @xcite that many of the functorial properties do hold if we restrict ourselves to normal spaces with kawamata log terminal ( klt ) singularities .
the following theorems summarises one of the main results in this direction .
[ thm : extpb ] let @xmath1 be a normal variety and @xmath13 a resolution of singularities .
assume that there exists a weil divisor @xmath14 on @xmath1 such that the pair @xmath15 is klt .
if @xmath16}_y \bigr ) = h^0 \bigl ( y_{\operatorname{reg}},\ , ^p_{y_{\operatorname{reg } } } \bigr)\ ] ] is any reflexive differential form on @xmath1 , then there exists a differential form @xmath17 that agrees on the complement of the @xmath5-exceptional set with the usual pull - back of the khler differential @xmath18 .
[ thm : pbgkkp ] let @xmath13 be a morphism of normal varieties .
assume that there exists a weil divisor @xmath14 on @xmath1 such that the pair @xmath15 is klt , and assume that the image of @xmath5 is not contained in the singular set @xmath6 .
then there exist pull - back morphisms of reflexive forms , @xmath19}_y ^{[p]}_x \quad \text{and } \quad { \ensuremath{d_\text{\rm refl}}}f : h^0 \bigl ( y,\ , ^{[p]}_y \bigr )
h^0 \bigl ( x,\ , ^{[p]}_x \bigr),\ ] ] that agrees with the usual pull - back of khler differentials wherever this makes sense .
theorems [ thm : extpb ] and its corollary , theorem [ thm : pbgkkp ] , allow to study reflexive differentials in the context of the minimal model program .
these results have been applied to a variety of settings , including a study of hyperbolicity of moduli spaces , @xcite or see @xcite for a survey , a partial generalisation of the beauville
bogomolov decomposition theorem @xcite and deformations of calabi
yau varieties @xcite .
the main result asserts the existence of a useful pull - back morphism in a more general setting .
[ thm : main ] let @xmath0 be any morphism between normal complex varieties .
assume that there exists a weil divisor @xmath14 on @xmath1 such that the pair @xmath15 is klt .
then there exists a pull - back morphism @xmath19}_y ^{[p]}_x,\ ] ] uniquely determined by natural universal properties .
the `` natural universal properties '' mentioned in theorem [ thm : main ] are a little awkward to formulate .
precise statements are given in theorem [ thm : pb - thma ] and section [ ssec : mwad ] .
in essence , it is required that the pull - back morphisms agree with the pull - back of khler differentials wherever this makes sense , and that they satisfy the composition law .
it should be noted that theorem [ thm : main ] does not require the image of @xmath5 to intersect the smooth locus of @xmath20 .
one particularly relevant setting to which theorem [ thm : main ] applies is that of a klt space @xmath1 , and the inclusion ( or normalisation ) of the singular locus , say @xmath21 .
it might seem surprising that a pull - back morphism exists in this context , because reflexive differential forms on @xmath1 are , by definition , differential forms defined on the _ complement _ of @xmath6 , and no boundary conditions are imposed that would govern the behaviour of those forms near the singularities .
in essence , theorem [ thm : main ] asserts that differential forms defined on the complement of @xmath6 determine forms _ on _ @xmath6 .
the following immediate corollary gives a precise formulation .
[ cor : singstrat ] let @xmath1 be a normal complex variety with klt singularities . define a stratification of @xmath1 by disjoint , locally closed , smooth subvarieties , @xmath22 , as follows .
consider the chain of closed subvarieties , @xmath23 given any index @xmath24 , let @xmath25 be the smooth locus of @xmath26 , that is , @xmath27 .
then , given any differential form @xmath28 we obtain a sequence of induced forms @xmath29 , for all @xmath30 .
the singularity stratification defined in corollary [ cor : singstrat ] is the coarsest stratification whose strata are locally closed and smooth .
the conclusion also holds for finer stratifications , such as whitney stratifications used in the discussion of intersection homology and perverse sheaves .
the following two examples show that it is generally not possible to construct reasonable pull - back morphisms for reflexive differentials on even the simplest log canonical spaces .
[ setting : ex12 ] let @xmath31 be a smooth elliptic curve , and let @xmath32 be the affine cone over @xmath33 .
an elementary computation shows that @xmath3 is normal and has an isolated , gorenstein , log canonical singularity at the vertex point @xmath34 .
the canonical bundle @xmath35}_x$ ] is in fact even trivial .
we denote the blow - up of the vertex point @xmath36 by @xmath37 .
the variety @xmath38 is then smooth .
denoting the @xmath39-exceptional curve by @xmath33 , there exists an isomorphism @xmath40 . observing that @xmath38 is isomorphic to the total space of the line bundle @xmath41
, we have constructed the following commutative diagram of morphisms between normal varieties , @xmath42^{_e \text { , inclusion } } \ar[d]_{c \text { , constant } } & & & { \widetilde}x \ar[rrr]^{\text{$$ , $ { \mathbb{a}}^1$-bundle } } \ar[d]^{\text{$$ , blow - up } } & & & c \\ \{x \ } \ar[rrr]_{_x \text { , inclusion } } & & & x. } \ ] ] [ ex : nlc1 ] we maintain assumptions and notation of setting [ setting : ex12 ] .
to give a pull - back map that agrees outside of @xmath33 with the usual pull - back map for khler differentials , it is then equivalent to give a sheaf morphism @xmath43 that is isomorphic away from @xmath33 .
such a morphism does not exist . if @xmath44 denotes a global generator of the canonical sheaf
, there is no section @xmath45 that agrees with @xmath2 away from the exceptional set @xmath33 .
[ ex : nlc2 ] we maintain assumptions and notation of setting [ setting : ex12 ] .
we consider reflexive one - forms on @xmath3 and assume there were pull - back morphisms @xmath46}_{x } \bigr ) h^0\bigl ( { \widetilde}x,\ , ^1_{{\widetilde}x } \bigr ) \quad \text{and } \quad { \ensuremath{d_\text{\rm refl}}}_x : h^0\bigl ( x,\ , ^{[1]}_{x } \bigr ) \underbrace{h^0\bigl ( \{x\ } , \ , ^1_{\{x\ } } \bigr)}_{= \ { 0\}}\ ] ] satisfying the following natural compatibility conditions . 1
. [ il:1 ] away from the singular point , the map @xmath47 agrees with the usual pull - back map of khler differentials .
[ il:2 ] the composition law holds .
in particular , diagram will imply that @xmath48 where @xmath49 denotes the usual pull - back of khler differentials .
since @xmath50 is the constant map , this implies that @xmath51 .
let @xmath52 be any non - vanishing differential form on the elliptic curve @xmath53 , and let @xmath54 be its pull - back to @xmath38 .
since @xmath55 and @xmath56 are isomorphic , the form @xmath57 induces a ( reflexive ) differential form on @xmath3 , say @xmath58}_{x }
\bigr ) = h^0\bigl ( x_{\operatorname{reg}},\ , ^1_{x_{\operatorname{reg } } } \bigr).\ ] ] property ( [ ex : nlc2].[il:1 ] ) then implies that @xmath59 , and @xmath60 this clearly contradicts property ( [ ex : nlc2].[il:2 ] ) , showing that pull - back morphisms satisfying these compatibility conditions can not exist . in the setting of example [ ex : nlc1 ] , there does exist a differential form with logarithmic poles along @xmath33 , say @xmath61 , that agrees with @xmath2 away from the singular set .
one could argue that @xmath62 should be taken as a pull - back of @xmath2 . while this might be a viable definition when discussing the blow - up morphism @xmath39
, problems occur as soon as one wishes to pull - back @xmath2 via the composition @xmath63 .
in addition to the problems originating from the existence of poles , example [ ex : nlc2 ] shows that there are other and more fundamental reasons why reasonable pull - back maps can not be defined for log canonical varieties : there is in general no way to define a pull - back map in a way that is compatible with the usual composition law .
the proof of our main result is technically somewhat involved .
the main idea , however , is quite elementary and straightforward .
consider the following simple setting .
[ setting : ex11 ] let @xmath1 be a variety with klt singularities .
assume that the singular locus is a smooth curve @xmath53 and that the singularities of @xmath1 are resolved by a single blow - up of the curve @xmath53 , say @xmath64 .
assume further that the exceptional set @xmath33 of this resolution map is irreducible and smooth over @xmath53 .
let @xmath65}_y \bigr)$ ] be any given reflexive differential form on @xmath1 .
theorem [ thm : extpb ] asserts the existence of a regular differential form @xmath66 that agrees outside of @xmath33 with the form @xmath2 .
[ ex : pb1 ] in setting [ setting : ex11 ] , define @xmath62 as the pull - back of the form @xmath2 to @xmath67 .
this choice is unique if we require our pull - back form to agree on the smooth locus with the usual pull - back of khler differentials .
[ ex : pb2 ] in setting [ setting : ex11 ] , let @xmath3 be a smooth variety and @xmath0 a morphism whose image is contained in @xmath53 .
we aim to define a pull - back form @xmath68 . a fundamental theorem of hacon and mckernan , ( * ?
1.5 ) , asserts that the fibres of @xmath69 are rationally connected manifolds . recalling that rationally connected manifolds do not admit non - trivial differential forms , the long exact sequence of the relative differential sequence
, @xmath70 shows that the restriction of @xmath62 to @xmath33 really is the pull - back of a form @xmath71 on @xmath53 .
let @xmath72 be the standard pull - back of the khler differential @xmath73 to @xmath3 .
the choice of @xmath72 is unique if we require that the pull - back morphisms satisfy the composition law . to this end , recall graber
starr s generalisation of tsen s theorem , ( * ? ? ?
1.1 ) , which gives the existence of a section @xmath74 , forming a commutative diagram @xmath75^{_e , \text { inclusion } } & & { \widetilde}y \ar[d]^{ , \text { resolution } } \\ x \ar[rr]^{f_c } \ar@/_3mm/[rrrr]_f & & c \ar[u]^{s } \ar@{^(->}[rr]^{_c , \text { inclusion } } & & y } \ ] ] choosing one such @xmath76 , observe that @xmath77 note that this completely dictates the choice of @xmath72 .
two problems occur when trying to adapt the ideas outlined in examples [ ex : pb1][ex : pb2 ] to the general setting , where @xmath3 is allowed to have arbitrary klt singularities .
[ prob:1 ] the exceptional set of a resolution morphism need not be irreducible , or relatively smooth over the singular locus .
fibres of resolution maps are known to be rationally chain connected , but khler differentials might well exist on these non - normal spaces .
[ prop:2 ] the result of graber harris
starr is specific to families over 1-dimensional base varieties , and is not known to hold in higher dimensions . to overcome the problem [ prob:1 ]
, we need to consider the sheaves @xmath78 of `` khler differentials modulo torsion '' and discuss their properties on reduced , reducible , and not necessarily normal schemes .
this is done in part [ part:1 ] .
there , we establish a number of fundamental universal properties and show that reduced , reducible , rationally chain connected schemes with simple normal crossings do not admit any `` khler differential modulo torsion '' .
the notions of torsion and torsion - free sheaves on reducible spaces do not seem be discussed much in the literature . for the reader s convenience , we recall the definition and establish basic properties in appendix [ app : a ]
. the problem [ prop:2 ] does not pose fundamental difficulties .
however , it does make the proof of our main theorem , given in part [ part:2 ] of this paper , somewhat awkward and lengthy as we constantly need to switch between the spaces in question and suitable coverings , for which a section @xmath76 exists . throughout the text
, we work over the complex number field . we discuss several notions of differential forms on singular spaces , and in each case define pull - back morphisms . to avoid the obvious potential for confusion ,
we clearly distinguish between the various notions of pulling back .
if @xmath0 is any morphism between varieties , we denote the sheaves of khler differentials by @xmath7 and @xmath79 , respectively .
the standard pull - back maps of khler differentials are denoted as @xmath80 part [ part:1 ] discusses the sheaf of khler differentials modulo torsion , traditionally denoted as @xmath81 . as we will see in section [ sec : tfd ] , there exists a meaningful notion of pulling back sections is this sheaf .
the associated morphisms of sheaves and vector spaces are denoted as @xmath82 part [ part:2 ] discusses reflexive differentials .
the sheaf of reflexive differentials will always be denoted by @xmath9}_x : = \bigl ( ^p_x \bigr)^{**}$ ] . following the notation introduced in theorem [ thm : pbgkkp ] , pull - back morphisms for reflexive differentials
will be denoted by @xmath83 .
[ not : sdos ] let @xmath3 be a scheme and @xmath84 a subscheme , with associated inclusion map @xmath85 .
sheaves @xmath86 on @xmath1 will often be viewed as sheaves on the ambient space @xmath3 .
if no confusion is likely to arise , we follow standard notation and write @xmath86 as a shorthand for the technically more correct @xmath87 .
the author would like to thank the members of the algebraic and arithmetic geometry groups at the university of freiburg for many inspiring discussions .
example [ ex : nlc2 ] was perceived in discussion with patrick graf .
fritz hrmann suggested a simplified proof of theorem [ prop : pull - back - tfd1 ] .
fred rohrer ( tbingen ) and angelo vistoli ( bologna ) kindly answered questions and provided references .
after the paper appeared in preprint form , we have learned from clemens jrder that parts of section [ sec : tfd ] overlap with @xcite , which discusses torsion - free differentials in a related settung , but using rather different language .
as indicated in the introduction , we need to discuss sheaves of `` khler differentials modulo torsion '' . the following notation will be used .
we refer to appendix [ app : a ] for a discussion of torsion sheaves on reduced , but possibly reducible spaces .
let @xmath88 be a morphism of reduced , quasi - projective schemes .
given any number @xmath89 , define @xmath90 as the cokernel of the sequence @xmath91 & \operatorname{tor}^p_{x / y } \ar[r]^{_{x / y } } & ^p_{x / y } \ar[r]^{_{x / y } } & \widecheck{}^p_{x / y } \ar[r ] & 0 } \ ] ] if @xmath1 is a point , we follow the usual notation and write @xmath92 instead of @xmath90 .
consider the torsion subsheaf @xmath93 , as introduced in definition .
sections in @xmath94 and @xmath95 are called ( relative ) _ torsion differentials_. by slight abuse of language , we refer to sections in @xmath90 and @xmath92 as ( relative ) _ torsion - free differentials_. [ gen : chartfd ] torsion differentials are characterised among khler differentials as those that vanish at general points of all irreducible components of @xmath3 . a torsion - free differential on @xmath3 vanishes if and only if it vanishes generically on all irreducible components of @xmath3 .
we refer to explanation [ expl : torgenpts ] for further discussion .
sequence is obviously of great importance in the discussion of torsion - free differentials .
we will therefore maintain the meaning of the symbols @xmath96 and @xmath97 throughout the present section [ sec : tfd ] .
again , if @xmath1 is a point , we write @xmath98 instead of @xmath96 . given a morphism between two varieties , we aim to show that the usual pull - back map of khler differentials always induces a pull - back map of torsion - free differentials , even if the image of the morphism is contained in the singular set of the target variety . the following proposition , which asserts that the pull - back of a torsion - differential is always torsion , is a first step in this direction . [ prop : pull - back - tfd1 ] let @xmath0 be a morphism of reduced , quasi - projective schemes .
if @xmath99 is a torsion - form on @xmath1 , then @xmath100 is a torsion form on @xmath3 .
in other words , @xmath101 the assertion of proposition [ prop : pull - back - tfd1 ] is clearly true if @xmath3 is irreducible and @xmath102 intersects the smooth locus of @xmath1 .
we claim that proposition [ prop : pull - back - tfd1 ] also holds in cases where @xmath102 is entirely contained in the singular locus .
one particularly relevant case is the inclusion of the singular locus , say @xmath103 .
if @xmath6 is itself smooth , then proposition [ prop : pull - back - tfd1 ] asserts that the pull - back of any torsion differential vanishes .
a simple case of this configuration is shown in figure .
@xmath104 the figure sketches a special case of proposition [ prop : pull - back - tfd1 ] . here
, it can be shown by elementary computation that the restriction of any torsion differential on @xmath3 to the singular locus @xmath105 vanishes . to be more precise , if @xmath106 denotes the inclusion map , then @xmath107 .
the proof of proposition [ prop : pull - back - tfd1 ] relies on the existence of a desingularisation map for which no analogue exists in the relative setting . as a matter of fact ,
proposition [ prop : pull - back - tfd1 ] becomes wrong when working with relative differentials , unless one makes rather strong additional assumptions . for a simple example , consider a sequence of morphisms , @xmath108^{f}_{\text{inclusion of except
. curve } } & & & & y \ar[rrrr]^{}_{\text{blow - up of smooth surface } } & & & & z. } \ ] ] a simple computation shows that @xmath109 . in particular , if @xmath110 is any open set and @xmath111 any non - vanishing differential on @xmath112 , then @xmath2 is torsion on @xmath112 , but @xmath113 is a torsion - free differential on the curve @xmath114 . before giving a proof of proposition [ prop : pull - back - tfd1 ] in section [ ssec : pfpbtfd1 ] , we note a number of corollaries which will be relevant in the further discussion . among these , the existence of a pull - back map for torsion - free differentials is the most important .
[ cor : pull - back - tfd1 ] let @xmath0 be a morphism of reduced , quasi - projective schemes . then there
exist unique morphisms @xmath115 and @xmath116 such that the following diagram with exact rows becomes commutative @xmath117^{f^ * _{y } } \ar[d]_{{\ensuremath{d_\text{\rm tor}}}f } & & f^ * ^p_{y } \ar[rr]^{f^ * _{y } } \ar[d]_{{\ensuremath{d_\text{\rm khler}}}f } & & f^ * \widecheck{}^p_{y } \ar[r ] \ar[d]^{{\ensuremath{d_\text{\rm tfree } } } } & 0 \\ 0 \ar[r ] & \operatorname{tor}^p_{x } \ar[rr]_{_{x } } & & ^p_{x } \ar[rr]_{_{x } } & & \widecheck{}^p_{x } \ar[r ] & 0 . } \end{split}\ ] ] proposition [ prop : pull - back - tfd1 ] immediately implies that the restricted morphism @xmath118 factorises via the torsion subsheaf @xmath119 ; this is clearly the unique morphism @xmath115 that makes the left square of commutative .
existence and uniqueness of @xmath116 follows from surjectivity of @xmath120 once we note that @xmath121 this finishes the proof of corollary [ cor : pull - back - tfd1 ] .
the morphism @xmath120 in is surjective .
this has a number of immediate consequences .
we mention two which will be relevant in the further discussion .
[ lem : dandcheckd ] in the setting of corollary [ cor : pull - back - tfd1 ] , if @xmath34 is any point such where @xmath122 vanishes , then @xmath116 vanishes as well . in other words ,
@xmath123 [ lem : cltfd ] in the setting of corollary [ cor : pull - back - tfd1 ] , let @xmath124 be any morphism from a reduced , quasi - projective scheme @xmath125 .
then @xmath126 consider the commutative diagram of sheaf morphisms , @xmath127^{g^ * ( { \ensuremath{d_\text{\rm khler}}}f ) } \ar[d]_{g^ * f^ * _{y } } & & g^ * ^p_{x } \ar[rr]^{{\ensuremath{d_\text{\rm khler}}}g } \ar[d]^{g^ * _{x } } & & ^p_{x ' } \ar[d]^{_{x ' } } \\ g^ * f^ * \widecheck{}^p_{y } \ar[rr]_{g^ * ( { \ensuremath{d_\text{\rm tfree}}}f ) } & & g^ * \widecheck{}^p_{x } \ar[rr]_{{\ensuremath{d_\text{\rm tfree}}}g } & & \widecheck{}^p_{x ' } } \end{split}\ ] ] obtained by composing the right commutative square of diagram for the morphism @xmath128 with the @xmath128-pull - back of the commutative square for the morphism @xmath5 . using the composition law for the pull - back of khler differentials , @xmath129 the outer square of is thus written as @xmath130^{{\ensuremath{d_\text{\rm khler}}}(f g ) } \ar[d]_{g^ * f^ * _{y } } & & & ^p_{x ' } \ar[d]^{_{x ' } } \\
g^ * f^ * \widecheck{}^p_{y } \ar[rrr]_{({\ensuremath{d_\text{\rm tfree}}}g ) g^ * ( { \ensuremath{d_\text{\rm tfree}}}f ) } & & & \widecheck{}^p_{x ' } } \end{split}\ ] ] as the pull - back of a surjective sheaf morphism , the map @xmath131 is surjective itself .
a comparison of diagram with the right square of diagram for the composed morphism @xmath132 thus immediately shows equation , as claimed . in the setting of corollary [ cor :
pull - back - tfd1 ] , the sheaf morphism @xmath116 induces a morphism between vector spaces of globally defined torsion - free forms , which we will again denote by @xmath133 . in the setting of corollary [ cor :
pull - back - tfd1 ] , given given a further morphism @xmath124 , lemma [ lem : cltfd ] implies that the following diagram is commutative , @xmath134_{{\ensuremath{d_\text{\rm tfree}}}f } \ar@/^5mm/[rrrr]^{{\ensuremath{d_\text{\rm tfree}}}(f g ) } & & h^0 \bigl(x,\ , \widecheck{}^p_x \bigr ) \ar[rr]_{{\ensuremath{d_\text{\rm tfree}}}g } & & h^0 \bigl(x',\ , \widecheck{}^p_{x ' } \bigr ) . } \ ] ] we are grateful to fritz hrmann for help with the following proof .
we maintain notation and assumptions of proposition [ prop : pull - back - tfd1 ] and assume that @xmath2 is a torsion form on @xmath1 .
recall from remark [ gen : chartfd ] or explanation [ expl : torgenpts ] that to show that @xmath113 is a torsion differential , it suffices to show that it vanishes at general points of each irreducible component of @xmath3 .
this simple observation has two important consequences .
first consequence : considering one component of @xmath3 at a time , we are free to assume without loss of generality that @xmath3 is irreducible . replacing @xmath3 by a suitable dense open subset
, we can even assume the following .
[ ass:211 ] the scheme @xmath3 is irreducible and smooth .
second consequence : let @xmath135 be an irreducible component that contains the image @xmath102 .
the morphism @xmath5 factors via the inclusion map , @xmath136_{f_0 } \ar@/^3mm/[rrrr]^{f } & & y_0 \ar[rr]_{_0 } & & y , } \ ] ] and @xmath137 .
since @xmath2 is torsion , @xmath138 vanishes at general points of @xmath139 , and is therefore a torsion form on @xmath139 .
proposition [ prop : pull - back - tfd1 ] will therefore follow for our given morphism @xmath5 if we can prove it for @xmath140 .
we can therefore assume without loss of generality that the following holds .
[ ass:212 ] the scheme @xmath1 is irreducible .
let @xmath141 be a resolution of singularities which exists because @xmath1 is irreducible .
choose a component of @xmath142 that surjects onto @xmath3 , and let @xmath38 be a desingularisation of that component .
we obtain a commutative diagram , @xmath143^{{\widetilde}f } \ar[d]_{_x } & { \widetilde}y \ar[d]^{_y } \\ x \ar[r]_{f } & y. } \ ] ] since @xmath2 vanishes at general points of @xmath1 and since @xmath144 is birational , it is clear that the pull - back @xmath145 vanishes at general points of @xmath67 .
but since @xmath67 is smooth by assumption [ ass:211 ] , this means that it vanishes everywhere , @xmath146 since @xmath147 is torsion - free and @xmath148 is generically smooth , we see that the pull - back morphism @xmath149 of globally defined forms is in fact injective .
it follows that @xmath150 if and only if @xmath151 .
this last pull - back is easily computed to be zero , @xmath152 this ends the proof of proposition [ prop : pull - back - tfd1 ] .
in the course of the proof of our main theorem , we will frequently need to consider klt spaces @xmath3 , strong resolution maps @xmath153 , and discuss torsion - free differentials on the exceptional set @xmath33 , which is an snc divisor embedded into the smooth space @xmath38 .
this section gathers several elementary facts about the sheaf @xmath154 which are needed in the discussion .
most of the material here will be known to experts .
we have nonetheless chosen to include full proofs for lack of an adequate reference .
to fix notation used later , we recall the definition and basic properties of snc pairs .
[ def : everythinglog2a ] let @xmath3 be a normal , quasi - projective variety and @xmath14 an effective weil divisor on @xmath3 .
given a point @xmath34 , we say that the pair @xmath155 is _ snc at @xmath36 _ if there exists a zariski - open neighbourhood @xmath112 of @xmath36 such that @xmath112 is smooth and such that @xmath156 is either empty , or a divisor with simple normal crossings . the pair @xmath155 is called _ snc _ if it is snc at every point of @xmath3 .
given a pair @xmath157 , let @xmath158 be the maximal open set of @xmath3 where @xmath157 is snc , and let @xmath159 be its complement , with the induced reduced subscheme structure .
the following notation and remark can be used to give an alternate definition of `` snc pair '' .
this will be used later to define snc in the relative setting .
[ not : di ] let @xmath157 be a pair , where the boundary divisor @xmath14 is written as a sum of its irreducible components @xmath160 .
if @xmath161 is any non - empty subset , we consider the scheme - theoretic intersection @xmath162 . if @xmath163 is empty , set @xmath164 .
[ rem : descrsnc ] in the setup of notation [ not : di ] , it is clear that the pair @xmath157 is snc if and only if all @xmath165 are smooth and of codimension equal to the number of defining equations , @xmath166 for all @xmath163 where @xmath167 .
the notion of relatively snc divisors has been used in the literature , but its definition has not been discussed much . for the reader s convenience , we reproduce the definition given in ( * ? ? ?
[ def : sncmorphism ] if @xmath157 is an snc pair and @xmath168 a surjective morphism to a smooth variety , we say that @xmath14 is _ relatively snc _ , or that @xmath169 is _ an snc morphism of the pair _
@xmath157 if for any set @xmath163 with @xmath170 all restricted morphisms @xmath171 are smooth of relative dimension @xmath172 .
[ rem : sncmor ] in the setting of definition [ def : sncmorphism ] , assume that @xmath168 is an snc morphism of the pair @xmath157 . then , since smooth morphisms are open , each of the sets @xmath165 dominates @xmath3 .
if @xmath173 is any point , set @xmath174 and @xmath175 . then @xmath176 is smooth and @xmath177 is an snc pair .
[ ex : sncmor ] the morphism @xmath178 , shown in figure is a prototypical example of an snc morphism .
if @xmath157 is an snc pair , then the identity map @xmath179 is an snc morphism if and only if @xmath180 . again assuming that @xmath157 is an snc pair , the constant map from @xmath3 to a point is always an snc morphism .
assume we are given an pair @xmath155 and a point @xmath34 such that @xmath155 is snc at @xmath36 .
it is well - known that there exists a neighbourhood @xmath181 , open in the analytic topology , and holomorphic coordinates @xmath182 such that @xmath183 , for a suitable number @xmath184 .
the following is the relative analogue of this fact .
[ lemdef : apdapt ] let @xmath157 be an snc pair and @xmath168 an snc morphism of the pair @xmath157 .
if @xmath34 is any point , then there exist neighbourhoods @xmath185 and @xmath186 open in the analytic topology , and holomorphic coordinates @xmath187 , @xmath188 , centred about @xmath36 respectively @xmath189 , and a number @xmath190 such that the following holds . 1 .
we have @xmath191 for all indices @xmath192 , and 2 .
the support of @xmath14 is given as @xmath193 . in the setting
above , we call the coordinates @xmath194 and @xmath195 _ adapted coordinates for the snc morphism @xmath169_. given an snc pair @xmath157 , the following two propositions describe the sheaf of torsion - free differentials on @xmath14 by relating them to the sheaves of differentials on each component of @xmath14 and to logarithmic differentials on @xmath3 , respectively .
these descriptions will later be used in the discussion of relative torsion - free differentials .
[ prop : namitfd0 ] let @xmath157 be an snc pair where @xmath14 is reduced , with irreducible components @xmath196 .
let @xmath197 be a morphism such that @xmath14 is relatively snc over @xmath1 .
given any number @xmath89 , consider the inclusion maps @xmath198 and the natural morphisms @xmath199 then the image of @xmath200 is naturally isomorphic to the sheaf of torsion - free differentials on @xmath14 , that is , @xmath201 .
recall from proposition [ prop : pftf ] that the push - forward sheaves @xmath202 are torsion - free sheaves on the reducible space @xmath14 . since subsheaves of torsion - free sheaves are torsion - free , corollary [ cor : sstf ] , it follows that @xmath203 is itself torsion - free .
the universal property of torsion - freeness , proposition [ prop : uptf ] , therefore gives a factorisation of @xmath200 as follows : @xmath204^{ } \ar@{->>}[rr ] & & \widecheck{}^p_{d / y } \ar@{->>}[rr]_(.4){\widecheck{ } } & & \operatorname{image}( ) \ar@{^(->}[r ] & \bigoplus_i ( _i ) _ * ^p_{d_i / y}. } \ ] ] the finish the proof , it suffices to show that @xmath205 is injective .
that , however , follows from proposition [ prop : ijtf ] because @xmath205 is generically injective : as a matter of fact , both @xmath200 and @xmath205 are isomorphic away from the singular set of @xmath14 . in the absolute case , the following statement appears without proof in @xcite .
[ prop : namitfd ] in the setup of proposition [ prop : namitfd0 ] , there exists a short exact sequence of sequence should be seen as a sheaf on @xmath3 , cf .
notation . ]
viewing @xmath207 and @xmath208 as sheaves on @xmath3 , proposition [ prop : namitfd0 ] asserts that to prove proposition [ prop : namitfd ] , it suffices to show that @xmath209 is exactly the kernel of the composed map @xmath210_r \ar@/^4mm/[rr]^{ } & ^p_{d / y } \ar[r]_(.4){ } & \bigoplus_i ^p_{d_i / y}. } \ ] ] setting @xmath211 , the kernel of @xmath212 is then this intersection of the kernels of all the @xmath213 , that is , @xmath214 the kernels of the @xmath213 are well understood . as a matter of fact ,
an elementary computation in adapted coordinates shows is also found in ( * ? ? ?
2.3(c ) on p. 13 ) . ]
that @xmath215 in particular , we see that @xmath216 are locally free subsheaves of @xmath217 .
since the intersection of two reflexive subsheaves is reflexive , equations and show that @xmath218 is a reflexive sheaf on @xmath3 , and isomorphic to @xmath219 on the open set @xmath220 .
since @xmath221 , and since two reflexive sheaves agree if and only if they agree on the complement of a small set , we obtain that @xmath222 on all of @xmath3 , thus finishing the proof of proposition [ prop : namitfd ] . given a smooth morphism @xmath88 , it is well - known that the restriction of @xmath217 to any fibre @xmath223 equals @xmath224 . using the descriptions given above , we show that the same holds for torsion - free differentials in the relative snc setting .
[ cor : ntorrest ] in the setup of proposition [ prop : namitfd0 ] , let @xmath225 be a smooth point of the image @xmath226 , and consider the preimage @xmath227 . then @xmath228 set @xmath229 and @xmath230 , and recall from remark [ rem : sncmor ] that @xmath231 is an snc pair . observe that to prove corollary [ cor : ntorrest ] , it suffices to show equation in the neighbourhood of any given point @xmath232 .
given one such @xmath36 , choose open neighbourhoods @xmath181 and @xmath233 and adapted coordinates @xmath234 and @xmath235 as in lemma and definition [ lemdef : apdapt ] . using these coordinates
, an elementary computation shows that the restriction of sequence to the fibre @xmath236 stays exact , and that the terms of the restricted sequence are identified as follows , @xmath237 recall from example [ ex : sncmor ] that the constant map from @xmath236 to a point is an snc morphism of the pair @xmath231 .
an application of proposition [ prop : namitfd ] therefore identifies the cokernel of sequence as @xmath238 , finishing the proof .
given a smooth morphism @xmath88 , the sequence of relative differentials on @xmath3 induces a canonical filtration on the sheaf @xmath7 .
the following proposition shows that the same statement holds for torsion - free differentials in the relative snc setting .
[ prop : tfdrelfilt ] let @xmath239 be an snc pair , where @xmath33 is a reduced divisor on @xmath3 .
let @xmath240 a morphism such that @xmath33 is relatively snc over @xmath1 .
if @xmath241 is any number , then there exists a filtration @xmath242 and for all @xmath243 sequences @xmath244 since @xmath169 is smooth , the sequence of relative differentials on @xmath3 is a short exact sequence of locally free sheaves on @xmath3 , @xmath245 following ( * ? ? ?
* ii.5 ex 5.16 ) , there exists an induced filtration @xmath246 and sequences for all @xmath243 , @xmath247 recalling sequence of proposition [ prop : namitfd ] , we define filtrations of @xmath248 and @xmath249 by setting @xmath250 with these definitions , it is clear that @xmath251 , for all indices @xmath252 .
we aim to understand the sheaves @xmath253 in more detail .
an explicit computation in adapted coordinates , which we leave to the reader , reveals two facts .
first , there is an isomorphism @xmath254 secondly , one obtains the description of the natural map between quotients given in the following commutative diagram @xmath255^(.55){q}_(.55){\text{natl .
map between quotients } } \ar[d]_{\cong } & & & \factor { \scr{f}}^r.{\scr{f}}^{r+1}. \ar[d]^{\cong } \\
^*^r_y \otimes \left ( ^{p - r}_{x / y}(\log e ) \otimes { \scr{j}}_e \right ) \ar[rrr]^(0.6){{\rm id}_{^*^r_y } \otimes a^{p - r}_{x / y } } & & & ^*^r_y \otimes ^{p - r}_{x / y}. } \end{split}\ ] ] in particular , since @xmath256 is injective and since @xmath257 is locally free on @xmath3 , it follows that @xmath258 is injective .
the snake lemma thus yields a commutative diagram of coherent sheaves on @xmath3 as follows , @xmath259 & { \scr{f}}^{r+1}(\log ) \ar[r ] \ar@{^{(}->}[d]^{a^p_{x / y}|_{{\scr{f}}^{r+1}(\log ) } } & { \scr{f}}^r(\log ) \ar[r ] \ar@{^{(}->}[d]^{a^p_{x / y}|_{{\scr{f}}^r(\log ) } } & \factor { \scr{f}}^r(\log).{\scr{f}}^{r+1}(\log ) .
\ar[r ] \ar@{^{(}->}[d]^{q } & 0\\ 0 \ar[r ] & { \scr{f}}^{r+1 } \ar[r ] \ar@{->>}[d ] & { \scr{f}}^r \ar[r ] \ar@{->>}[d ] & \factor { \scr{f}}^r.{\scr{f}}^{r+1}. \ar[r ] \ar@{->>}[d ] & 0\\ 0 \ar[r ] & \widecheck { \scr{f}}^{r+1 } \ar[r ] & \widecheck { \scr{f}}^r \ar[r ] & \factor \widecheck { \scr{f}}^r.\widecheck { \scr{f}}^{r+1}. \ar[r ] & 0.\\ } \end{split}\ ] ] in summary , we obtain the following description of the successive quotients in the filtration of @xmath154 , @xmath260 this finishes the proof of proposition [ prop : tfdrelfilt ] .
it is well - known that rationally chain connected compact manifolds do not admit any differential forms . here
, we show that the same holds for torsion - free differentials on rationally chain connected varieties with arbitrary singularities .
[ thm : diffrcspaces ] let @xmath3 be a reduced , projective scheme .
assume that @xmath3 is rationally chain connected .
then @xmath261 in theorem [ thm : diffrcspaces ] , we do not assume that @xmath3 is irreducible .
the statement of theorem [ thm : diffrcspaces ] becomes wrong if one replaces torsion - free differentials with khler differentials .
for an example , let @xmath262 be the union of two distinct lines in @xmath263 .
the reducible variety @xmath3 is then clearly rationally chain connected .
the sheaf @xmath264 of khler differentials contains a non - trivial torsion subsheaf , supported at the intersection point @xmath265 . as a skyscraper sheaf supported at a single point , @xmath266 , and hence @xmath264 ,
do have non - trivial sections .
similar examples exist where @xmath3 is irreducible and rationally connected .
the paper @xcite discusses cones where @xmath266 is non - trivial , and supported at the vertex point .
the statement of theorem [ thm : diffrcspaces ] becomes wrong if one replaces torsion - free differentials with reflexive differentials ; example discusses a non - trivial reflexive form on the ( rationally chain connected ) cone over an elliptic curve .
the main reason for failure theorem [ thm : diffrcspaces ] in this setting is the fact that @xmath3 has log canonical rather than klt singularities . for rationally chain connected spaces
@xmath3 with klt singularities , it is shown in ( * ? ? ?
5.1 ) that @xmath267}_x \bigr ) = 0 $ ] for all @xmath89 . for reflexive tensor operation other than @xmath268}$ ] , the question becomes rather subtle .
we refer to ( * ? ? ?
3 ) for a discussion of known facts and for examples .
the arguments used to prove of theorem [ thm : diffrcspaces ] follow the standard proof for the non - existence of khler differentials on rationally connected manifolds .
let @xmath269 be any reflexive differential . by remark [ gen : chartfd ]
, it suffices to show that @xmath2 vanishes at the general point of every irreducible component of @xmath3 . in other words ,
given any irreducible component @xmath270 with inclusion map @xmath271 , we need to show that @xmath272 .
the assumption that @xmath3 is rationally chain connected implies that there exists a fixed point @xmath273 , such that general points of @xmath274 can be connected to @xmath275 using a finite - length chain of rational curves .
the following is thus an immediate consequence .
[ cons : rcc ] there exists a nodal chain of smooth rational curves @xmath276 with marked points @xmath277 as as shown in figure [ fig : chain ] , a smooth , irreducible , quasi - projective variety @xmath278 and a morphism @xmath279 with the following properties . 1 .
[ il : bonny ] the variety @xmath280 dominates the component @xmath274 .
[ il : clyde ] the restriction of @xmath281 to @xmath282 is constant .
( 0,0 ) to [ out=35 , in=145 ] ( 2.5,0 ) ; ( 1.25,0.7 ) node @xmath283 ; ( 2,0 ) to [ out=35 , in=145 ] ( 4.5,0 ) ; ( 3.25,0.7 ) node @xmath284 ; ( 6,0 ) to [ out=35 , in=145 ] ( 8.5,0 ) ; ( 7.25,0.7 ) node @xmath285 ; ( 8,0 ) to [ out=35 , in=145 ] ( 10.5,0 ) ; ( 9.25,0.7 ) node @xmath286 ; ( 2.25,0.15 ) circle ( 2pt ) ; ( 2.25,0 ) node[below ] @xmath287 ; ( 4.25,0.15 ) circle ( 2pt ) ; ( 4.25,0 ) node[below ] @xmath288 ; ( 6.25,0.15 ) circle ( 2pt ) ; ( 6.25,0 ) node[below ] @xmath289 ; ( 8.25,0.15 ) circle ( 2pt ) ; ( 8.25,0 ) node[below ] @xmath290 ; ( 10.25,0.15 ) circle ( 2pt ) ; ( 10.25,0 ) node[below ] @xmath291 ; ( 5.25,0.3 ) node[below ] ; with the notation introduced above , the following lemma is key to the proof of theorem [ thm : diffrcspaces ] .
[ lem : xp ] setup as above . for any index @xmath292
, we consider the following two restrictions of @xmath281 , @xmath293 then the following two equations hold for all indices @xmath292 @xmath294 assuming for a second that lemma [ lem : xp ] holds , consider equation for @xmath295 .
write @xmath296 as a composition @xmath297^{_{c_1,0 } } & & x_0 \ar[r]^{ } & x } \ ] ] equation and the composition law for pull - back of torsion - free differentials , lemma [ lem : cltfd ] give a vanishing of forms , @xmath298 recalling from ( [ cons : rcc].[il : bonny ] ) that @xmath299 is dominant , equation says that the restriction of @xmath2 to the component @xmath274 vanishes generically .
the required vanishing of @xmath300 then an immediate consequence .
theorem [ thm : diffrcspaces ] is thus a consequence of lemma [ lem : xp ] .
it remains to prove lemma [ lem : xp ] .
the proof proceeds by descending induction on @xmath301 . to this end
, the following statements will be shown in section [ sssec : s1][sssec : s3 ] below .
[ claim : xp1 ] equation holds for @xmath302 .
[ claim : xp2 ] for all indices @xmath292 , equation holds for the index @xmath301 if equation holds for @xmath301 .
[ claim : xp3 ] for all indices @xmath303 , equation holds for the index @xmath304 if equation holds for @xmath301 . for @xmath302 ,
equation follows from ( [ cons : rcc].[il : clyde ] ) , which asserts that @xmath305 is constant .
the pull - back map of khler differentials is thus zero , @xmath306 . by lemma [ lem : dandcheckd ] ,
the pull - back map of torsion - free differentials is therefore zero too , so that @xmath307 as claimed .
let @xmath292 be any given index , and assume that equation holds for @xmath301 .
the following morphisms are relevant in our discussion @xmath308_{ , \text { inclusion } } \ar@/^0.5cm/[rrrr]^{_{p_j } } & & h
c_j \ar[d]^{_h,\text { projection } } \ar[rr]_{_{c_j } } & & x\\ & & h } \ ] ] the product structure of @xmath309 immediately gives a splitting @xmath310 decompose @xmath311 accordingly as @xmath312 , we aim to show that both @xmath313 and @xmath314 are zero . first
, if @xmath315 is any fibre of @xmath316 , then @xmath317 is anti - ample .
it follows that @xmath318 only has the trivial section , so @xmath319 .
secondly , if follows from equation and from the composition law for pull - back of torsion - free differentials , lemma [ lem : cltfd ] , that @xmath320 we obtain that @xmath321 .
since @xmath322 is a trivial vector bundle , @xmath323 , this shows vanishing of @xmath313 . in summary
, we have seen that @xmath324 is zero , as asserted in claim [ claim : xp2 ] .
let @xmath303 be any given index , and assume that equation holds for the index @xmath301 .
since @xmath325 , we obtain a factorisation @xmath326_{,\text { inclusion } } \ar@/^0.5cm/[rrrr]^{_{p_{j-1 } } } & & h
c_j \ar[rr]_{_{c_j } } & & x } \ ] ] the composition law for the pull - back of torsion - free differentials , lemma [ lem : cltfd ] , then asserts that @xmath327 which immediately shows the desired vanishing .
this finishes the proof of claim [ claim : xp3 ] , and hence of theorem [ thm : diffrcspaces ] .
we aim to construct a pull - back map for reflexive differentials , which will turn out to be uniquely determined by universal properties . to formulate these properties ( `` composition law '' , `` compatibility with khler differentials '' ) in a technically correct manner
, it seems easiest to use the language of functors .
the following definition fixes the category .
let @xmath3 be a normal , irreducible variety .
we call @xmath3 a _ klt base space _ if there exists a @xmath328-divisor @xmath14 on @xmath3 such that the pair @xmath157 is klt . a morphism between klt base spaces
is simply a morphism of varieties .
the following theorem contains the main result of this paper . its proof is given in sections [ ssec : pfcpb][sec : pfpb ] , starting on page .
an extension of theorem [ thm : pb - thma ] to morphisms with arbitrary domain is discussed in section .
[ thm : pb - thma ] there exists a unique contravariant functor @xmath329}_x \bigr ) \end{matrix}\ ] ] that satisfies the following `` compatibility with khler differentials '' .
if @xmath330 is a morphism of klt base spaces such that the open set @xmath331 is not empty , then there exists a commutative diagram , @xmath332}_x \bigr ) \ar[rr]^{{\ensuremath{d_\text{\rm refl}}}f } \ar[d]_{\operatorname{restriction}_x } & & h^0 \bigl ( z,\ , ^{[p]}_z \ar[d]^{\operatorname{restriction}_z } \bigr ) \\
h^0 \bigl ( x_{\operatorname{reg}},\ , ^p_{x_{\operatorname{reg } } } \bigr ) \ar[rr]_{{\ensuremath{d_\text{\rm khler}}}(f|_{z } ) } & & h^0 \bigl ( z,\ , ^p_{z } \bigr ) , } \end{split}\ ] ] where @xmath333 denotes the usual pull - back of khler differentials , and @xmath83 denotes the linear map of complex vector spaces induced by the contravariant functor .
[ rem:43 ] in the setting discussed in diagram , we have equalities @xmath10}_x\bigl|_{x_{\operatorname{reg } } } = ^p_{x_{\operatorname{reg } } } \quad \text{and } \quad ^{[p]}_z\bigl|_{z } = ^p_{z}.\ ] ] this justifies the use of the word `` restriction '' in diagram . since @xmath3 and @xmath334 are normal , hence smooth in codimension one , the restriction maps @xmath335 and @xmath336 are clearly isomorphic .
consider a morphism @xmath330 of klt base spaces whose image is contained in the singular locus of @xmath3 . in this case , the set @xmath337 discussed in theorem [ thm : pb - thma ] is empty .
the compatibility condition formulated in the theorem is then also empty , that is , always satisfied .
this does not mean that @xmath83 is an arbitrary map .
the pull - back map @xmath83 is in fact uniquely defined by the functorial properties ( `` composition rule '' ) , and by the requirement that the pull - back maps of other , dominant , morphisms need to satisfy compatibility with khler differentials .
the proof of proposition [ prop : cmm ] will illustrate this principle . given a klt base space ,
the functorial formulation of theorem [ thm : pb - thma ] would in principle allow to write @xmath338 as a shorthand for the space @xmath267}_x \bigr)$ ] of reflexive differential forms on @xmath3 . in order to avoid confusion and incompatibility with the literature
, we will never use this notation .
the compatibility with khler differentials implies that practically all properties known from the pull - back map of khler differentials also hold in the reflexive setting .
we mention some of the more immediate examples in the present section [ ssec:5b ] .
[ prop : coi ] let @xmath3 be any klt base space , @xmath339 any open set and @xmath330 the inclusion map .
then @xmath340}_x \bigr ) h^0 \bigl ( z,\ , ^{[p]}_z \bigr)\ ] ] equals the standard restriction map .
define the subset @xmath337 as in theorem [ thm : pb - thma ] and observe that @xmath337 is not empty . using the isomorphisms @xmath335 and @xmath341
, the claim follows from the observation that @xmath333 is the standard restriction map .
[ prop : mtsmts ] let @xmath330 be any dominant morphism between klt base spaces , where @xmath3 is smooth .
then @xmath83 equals the composition of the following maps @xmath342}_x \bigr ) = h^0 \bigl ( x,\ , ^p_x \bigr ) \xrightarrow{\quad{\ensuremath{d_\text{\rm khler}}}\quad } h^0 \bigl ( z,\ , ^p_z \bigr ) \xrightarrow{\quad\varphi\quad } h^0 \bigl ( z,\ , ^{[p]}_z \bigr),\ ] ] where @xmath343 is induced by the standard map from the sheaf @xmath344 into its double dual .
[ prop : pbdm ] let @xmath330 be any dominant morphism between klt base spaces .
then the associated map @xmath340}_x
h^0 \bigl ( z,\ , ^{[p]}_z \bigr)\ ] ] is injective .
define the subset @xmath337 as in theorem [ thm : pb - thma ] and observe that @xmath337 is not empty , and that @xmath345 is dominant .
proposition [ prop : pbdm ] then follows immediately from injectivity of @xmath333 .
the following proposition [ prop : cmm ] and corollary [ cor : pbsheaf ] imply that the pull - back map for reflexive differentials is already defined on sheaf level , just as the pull - back map for khler differentials is .
[ prop : cmm ] let @xmath330 be any morphism between klt base spaces . then associated map @xmath340}_x \bigr )
h^0 \bigl ( z,\ , ^{[p]}_z \bigr)\ ] ] is a morphism of @xmath346-modules .
[ cor : pbsheaf ] if @xmath330 is any morphism between klt base spaces , then the pull - back morphism @xmath340}_x \bigr )
h^0 \bigl ( z,\ , ^{[p]}_z \bigr)\ ] ] is induced by a sheaf morphism @xmath19}_x ^{[p]}_z.\ ] ] immediate from compatibility with open immersions and compatibility with module multiplication , propositions [ prop : coi ] and [ prop : cmm ] .
the proof of proposition [ prop : cmm ] makes use of the following elementary construction , which we note for later reference .
[ cons : exv ] let @xmath330 be any morphism between normal varieties , and let @xmath347 be a resolution of singularities
. then there exists a smooth variety @xmath348 and a commutative diagram , @xmath349_{}^{\txt{\scriptsize resolution of\\\scriptsize singularities } } \\
v \ar@/^5mm/[rrrru]^{a } \ar[rr]_{g\text { , surjective } } & & z \ar[rr]_f & & x. } \end{split}\ ] ] one way of construction goes as follows .
choose a component @xmath350 that surjects onto @xmath334 , and let @xmath351 be a desingularisation .
let @xmath352 be the relative dimension of @xmath353 , and @xmath354 the intersection of @xmath352 general hyperplanes ( if @xmath355 , set @xmath356 ) .
the variety @xmath348 is then smooth , with natural morphisms to @xmath334 and @xmath38 making the diagram commutative .
the morphism to @xmath357 constructed in this manner is surjective and generically finite .
given reflexive forms @xmath358}_x \bigr)$ ] and a function @xmath359}_x \bigr)$ ] , we need to show that @xmath360 there are two situation where this is easily true . 1 .
[ il : xa ] if the target of @xmath5 is smooth , then proposition [ prop : mtsmts ] implies that holds because both @xmath49 and @xmath343 are @xmath346-linear .
[ il : xb ] if @xmath5 is dominant , then @xmath337 is non - empty , and equation holds because compatibility with khler differentials implies that it holds on the open set @xmath337 . proposition [ prop : cmm ] is hence shown for dominant morphisms , and for morphisms with smooth target . if @xmath5 is neither , apply construction [ cons : exv ] to obtain a commutative diagram as in .
we have seen above that @xmath361 is @xmath362-linear .
it is injective by proposition [ prop : pbdm ] . to prove
, it will therefore suffice to prove the analogous equation for the composed morphism @xmath363 , @xmath364 that , however , follows from functoriality , @xmath365 , because we have seen in above that @xmath366 and @xmath367 are linear over global sections of their respective structure sheaves . if @xmath3 is any normal variety , then to give a reflexive differential on @xmath3 it is equivalent to give a khler differential on the smooth locus @xmath368 . more precisely , if @xmath11 denotes the inclusion of the smooth locus into @xmath3 , then @xmath9}_x = _ * ^p_{x_{\operatorname{reg}}}$ ] .
this description of reflexive differentials immediately allows to define _ reflexive wedge products _ , that is , for all @xmath89 and @xmath258 morphisms @xmath369}_x \otimes_{{\scr{o}}_x } ^{[q]}_x & & ^{[p+q]}_x \\
\wedge : & h^0 \bigl ( x,\ , ^{[p]}_x \bigr ) \otimes_{{\scr{o}}_x(x ) } h^0 \bigl ( x,\ , ^{[q]}_x \bigr ) & & h^0 \bigl ( x,\ , ^{[p+q]}_x \bigr ) \end{array}\ ] ] that , on the smooth locus of @xmath3 , agree with the usual wedge products .
we define @xmath370-linear _ reflexive exterior derivatives _ in the same fashion , for all @xmath89 , @xmath371}_x & & ^{[p+1]}_x \\
d : & h^0 \bigl ( x,\ , ^{[p]}_x \bigr ) & & h^0 \bigl ( x,\ , ^{[p+1]}_x \bigr ) .
\end{array}\ ] ] in this setting , the proof of proposition [ prop : cmm ] applies almost verbatim to give the following compatibility result .
let @xmath330 be any morphism between normal varieties .
then the pull - back maps for reflexive differentials and sheaves of reflexive differentials commute with reflexive wedge products and reflexive exterior derivatives . the properties listed above allow for the construction of a meaningful pull - back map even for morphisms where only the target is assumed to be a klt base space . to this end , let @xmath330 be any morphism between normal varieties and assume that @xmath3 a klt base space . observing that the smooth locus @xmath372 is a klt base space as well , and recalling that the restriction map @xmath373}_z \bigr )
h^0 \bigl ( z_{\operatorname{reg}},\ , ^p_{z_{\operatorname{reg } } } \bigr)\ ] ] is isomorphic , define a pull - back map for reflexive differentials as the composition of the following two maps , @xmath342}_x \bigr ) \xrightarrow{\quad{\ensuremath{d_\text{\rm refl}}}(f|_{x_{\operatorname{reg } } } ) \quad } h^0 \bigl ( z_{\operatorname{reg}},\ , ^p_{z_{\operatorname{reg } } } \bigr ) \xrightarrow{\quad\operatorname{restriction}_z^{-1}\quad } h^0 \bigl ( z,\ , ^{[p]}_z \bigr).\ ] ] as there is no possibility of confusion , we denote this map again by @xmath83 .
is is easy to show that all properties listed in section [ ssec:5b ] also hold for this generalised map . in particular , @xmath83 is induced by a sheaf morphism .
we have thus constructed maps @xmath374}_x ^{[p]}_z$ ] and @xmath375}_x \bigr )
h^0 \bigl ( z,\ , ^{[p]}_z \bigr)$ ] for all indices @xmath89 .
given a morphism of klt base spaces and a reflexive form on the target space , the following proposition constructs a reflexive form on the domain that satisfies a weak universal property . [ prop : const ] let @xmath376 be any morphism of klt base spaces , and let @xmath377}_x \bigr)$ ] be any reflexive @xmath89-form on @xmath3 .
then there exists a unique reflexive @xmath89-form @xmath378}_z \bigr)$ ] satisfying the following universal property .
given any commutative diagram @xmath379_{}^{\txt{\scriptsize resolution of\\\scriptsize singularities } } \\
v \ar@/^5mm/[rrrru]^{a } \ar[rr]_{g\text { , surjective } } & & z_{\operatorname{reg } } \ar[rr]_{f|_{z_{\operatorname{reg } } } } & & x , } \end{split}\ ] ] where @xmath348 is smooth , let @xmath380 be the unique differential form on @xmath38 that agrees with @xmath2 wherever @xmath381 is isomorphic .
then @xmath382 the existence of the form @xmath62 used in the formulation of proposition [ prop : const ] is the main result of the paper @xcite .
we refer to ( * ? ? ?
* thm . 1.4 and rem . 1.5.2 ) for details . in analogy with remark [ rem:43 ] ,
we have used the canonical identification @xmath9}_z|_{z_{\operatorname{reg } } } \cong ^p_{z_{\operatorname{reg}}}$ ] implicitly in the formulation of equation .
[ rem : exv2 ] in the setting of proposition [ prop : const ] , if @xmath383 is not contained in the singular locus of @xmath3 , then it follows immediately from the universal property that @xmath73 is the unique reflexive form whose restriction to the open set @xmath384 satisfies @xmath385 .
[ rem : exv ] given an arbitrary resolution of singularities @xmath347 , construction [ cons : exv ] shows that there always exist smooth varieties @xmath348 and morphisms @xmath386 , @xmath128 as in diagram .
we will see in section [ ssec : popropconst ] that proposition [ prop : const ] is in fact a corollary of the following , seemingly weaker lemma .
[ lem : const ] let @xmath376 be any morphism of klt base spaces , and let @xmath377}_x \bigr)$ ] be any reflexive @xmath89-form on @xmath3 . then there exists a resolution of singularities @xmath347 and a reflexive @xmath89-form @xmath387}_z \bigr)$ ] satisfying a universal property similar to the one spelled out in proposition [ prop : const ] : given any smooth variety @xmath348 and morphisms @xmath386 , @xmath128 forming a commutative diagram as in , then equation holds .
we will later see that the form @xmath73 constructed in lemma [ lem : const ] for one specific resolution map will also work for any other .
the remainder of the present section [ ssec : preppf ] is devoted to the proof of lemma [ lem : const ] .
though not extremely involved on a conceptual level , the proof is somewhat lengthy to write down . to help the reader
maintain an overview , we have subdivided the proof into a number of relatively independent steps , given in sections [ sssec : pbanotation][sssec : polconstend ] below .
we maintain assumptions and notation of lemma [ lem : const ] throughout the proof . in order to construct a reflexive form on @xmath334
, it suffices to construct the form away from a set of codimension two .
since @xmath334 is a klt base space , hence normal , we can therefore assume without loss of generality that the following holds .
[ aass : smooth ] the space @xmath334 is smooth . in case where the image of @xmath5 is not contained in the singular set of @xmath3
, it has been shown in ( * ? ? ?
4.3 ) that there exists a differential form @xmath73 on @xmath334 that agrees on the open set @xmath388 with the usual pull - back of the khler differential @xmath389 .
the differential @xmath73 clearly satisfies all requirements stated in lemma [ lem : const ] , so that the proof is already finished in this case .
we will therefore assume without loss of generality that the following holds .
[ aass : sing ] the image of @xmath5 is contained in the singular set of @xmath3 . we define @xmath390 as the zariski closure of the image of @xmath5 , that is , @xmath391 .
choose a desingularisation @xmath153 with the additional property that the preimage @xmath392 has pure codimension one and forms a divisor with simple normal crossing support .
finally , let @xmath393 be the union of those components that dominate ( = surject onto ) the irreducible variety @xmath394 .
its irreducible components are denoted as @xmath395 one way to describe the geometry of @xmath3 near general points of @xmath394 is by looking at a family of sufficiently general complete intersections @xmath396 , and by studying the varieties @xmath397 at their intersection points with @xmath394 . at general points of @xmath394 ,
the family defines a morphism , and it is often notationally convenient to discuss the varieties @xmath397 as being fibres of that morphism .
this idea is not new , and is explained in great detail in ( * ? ? ?
the following lemma summarises the results and fixes notation used throughout the remainder of the proof of lemma [ lem : const ] .
figure [ fig : tswah ] , taken from the preprint version of the paper @xcite , illustrates the setup .
@xmath398{family_desing } } \put ( 3.6 , 3.5){\scriptsize divisor $ e_0 $ } \put ( 3.6 , 3.4){\vector(-1 , -1){0.4 } } \put ( 4.0 , 3.2){\scriptsize divisor $ e_1 $ } \put ( 4.0 , 3.1){\vector(-1 , -1){0.4 } } \end{picture } } \ar@<-.5mm>[rrr]^{_{}}_{\text{resolution map } } \ar@(d , l)[ddrrr]_(.6){_{ } } & & & * + + { \begin{picture}(4,4)(0,0 ) \put ( 0.0 , 4.2){singular space $ u_{}$ } \put ( 0.0 , 0.2){\includegraphics[height=3.5cm]{family_sing } } \put ( 3.8 , 1.3){\scriptsize curve $ t_{}$ } \put ( 3.8 , 1.5){\vector(-1 , 1){0.4 } } \end{picture } } \ar[dd]_(.7){_{}}^(.7){\text{projection to $ t_{}$ } } \\\\ & & & * + + { \begin{picture}(4,0.6)(0,0 ) \put ( 0.0 , 0.5){\includegraphics[width=3.5cm]{family_base } } \put ( 0.0 , 0.0){curve $ t_{}$ } \end{picture } } } \ ] ] the figure sketches the situation of lemma [ lem : proj ] in case where @xmath3 is a threefold and @xmath394 is a curve .
over @xmath399 , the exceptional set of the resolution map @xmath381 is a reducible divisor , @xmath400 .
the composed map @xmath178 is an snc morphism of the pair @xmath401 .
[ lem : proj ] in the setup of section [ sssec : pbanotation ] , there exists a dense , zariski - open set @xmath402 , an open covering of @xmath403 by sets @xmath404 that are open in the analytic topology , and commutative diagrams of holomorphic morphisms , @xmath405^(.4){{\widetilde}{}_{ } \text { , inclusion } } \ar@/_4mm/[lld]_{_{ } } \ar[d]_{|_{{\widetilde}u_{ } } } & & { \widetilde}x { \widetilde}x \ar@<4.5mm>[d]^{ } \ar@<-4.5mm>[d]_{|_{{\widetilde}x}}\\ t_{ } & & u_{ } \ar[ll]^{_{ } } \ar[rr]_(.4){_{ } \text { , inclusion } } & & x x } \ ] ] where @xmath406 such that the following extra conditions hold 1 .
[ il : qx1 ] the variety @xmath407 is smooth and not empty .
the sheaf @xmath408 is trivial . 2 .
[ il : qx2 ] setting @xmath409 , we have @xmath410 .
3 . [ il : qx3 ] if @xmath411 is any irreducible component , then @xmath412 is smooth .
il : qx4 ] the restrictions @xmath413 are the identity morphisms . 5 .
[ il : qx5 ] the holomorphic maps @xmath178 are smooth .
the divisors @xmath414 are relatively snc over @xmath399 .
the existence of an open set @xmath415 satisfying conditions ( [ lem : proj].[il : qx1])([lem : proj].[il : qx3 ] ) is clear . shrinking @xmath415 further , if necessary ,
the standard technique of `` projection to a subvariety '' , explained and proven in ( * ? ? ? * prop .
2.26 ) , yields the existence of a finite , tale covering map @xmath416 and a commutative diagram , @xmath417^(.4){{\widetilde}{}\text { , finite , tale } } \ar@/_4mm/[lld]_{ } \ar[d]_{p } & & & { \widetilde}x { \widetilde}x \ar@<4.5mm>[d]^{ } \ar@<-4.5mm>[d]_{|_{{\widetilde}x}}\\ b & & y \ar[ll]^{ } \ar[rrr]_(.4){\text { , finite , tale } } & & & x x , } \ ] ] where @xmath418 , where @xmath419 , and where @xmath89 and @xmath420 denote the obvious projections .
the morphism @xmath169 has the additional property that its restriction to @xmath421 is the identity map .
shrinking @xmath415 further , if necessary , generic smoothness of morphisms and its logarithmic analogue , ( * ? ? ?
. 2.11 ) , allow to assume that @xmath200 is smooth , and that the divisor @xmath422 is relatively snc over @xmath423 .
to end the proof , it suffices to find a covering of @xmath424 by analytically open sets @xmath425 that are small enough so that their preimages are disjoint unions of @xmath426-many open sets , @xmath427 each canonically identified with @xmath428 .
choosing one @xmath429 for each given set @xmath428 , the morphisms @xmath430 and @xmath178 are immediately obtained from these identifications .
conditions ( [ lem : proj].[il : qx4 ] ) and ( [ lem : proj].[il : qx5 ] ) follow from the properties of @xmath169 and @xmath200 .
we follow the ideas outlined in section [ sssec : idea ] .
given any point @xmath431 , a fundamental result of hacon - mckernan asserts that the fibre @xmath432 is rationally chain connected .
using the results obtained in section [ sec : tfdrcc ] , this implies that relative differentials in @xmath433 vanish modulo torsion when restricted to any component of @xmath434 .
this is a first indication of the principle that `` the restriction of any differential to @xmath435 comes from @xmath407 '' , as formulated and proven in the following lemma .
[ lem : rdwxc1 ] in the setup of section [ sssec : proj ] , the pull - back map of torsion - free differentials , @xmath436 is isomorphic .
choosing an open cover of @xmath407 as in lemma [ lem : proj ] , it suffices to show that the pull - back maps associated with the restricted morphisms , @xmath437 are isomorphisms , for all @xmath438 .
let @xmath438 be any given index .
the pull - back map @xmath439 is clearly injective . to prove surjectivity ,
consider the filtration and the sequences introduced in proposition [ prop : tfdrelfilt ] , @xmath440 which exist for all @xmath243 .
observe that in case where @xmath441 , sequence yields @xmath442 . to prove lemma [ lem : rdwxc1 ]
, we aim to show that @xmath443 since the vector bundles , @xmath444 are trivial by assertion ( [ lem : proj].[il : qx1 ] ) of lemma [ lem : proj ] , equation will follow from an inductive argument using the sequences once we show that @xmath445 using corollary [ cor : ntorrest ] to identify the restriction @xmath446 with @xmath447 for all points @xmath448 and all fibres @xmath449 , equation will in turn follow from the stronger claim that @xmath450 to prove , recall that @xmath415 is a klt base space .
a fundamental result of hacon - mckernan ( * ? ? ? * cor .
1.5(1 ) ) thus implies that fibres @xmath434 , being fibres of the birational resolution map @xmath381 , are rationally chain connected .
the vanishing asserted in is therefore an immediate consequence of the non - existence of torsion - free forms in rationally chain connected spaces , as asserted in theorem [ thm : diffrcspaces ] . using the standard fact that klt base spaces have rational singularities , it might be possible to give a proof of lemma [ lem : rdwxc1 ] using namikawa s analysis of mixed hodge structures , ( * ? ? ?
* lemma 1.2 ) , rather than the more elementary theorem [ thm : diffrcspaces ] in the setup of section [ sssec : proj ] , there exists a unique torsion - free form @xmath451 such that @xmath452 where @xmath453 denotes the obvious inclusion .
consider the non - empty , zariski - open set @xmath454 .
we obtain a torsion - free form @xmath455 the following elementary lemma summarises what we know about @xmath456 .
in essence , it shows that the form @xmath456 satisfies equation on the open set @xmath337 .
[ lem : bpx ] setting as above . given a smooth variety @xmath348 and morphisms @xmath386 , @xmath128 forming a commutative diagram as in of proposition [ prop : const ] , @xmath457_{}^{\txt{\scriptsize resolution of\\\scriptsize singularities } } \\
v \ar@/^5mm/[rrrru]^{a } \ar[rr]_{g\text { , surjective } } & & z_{\operatorname{reg } } \ar[rr]_{f|_{z_{\operatorname{reg } } } } & & x , } \end{split}\ ] ] set @xmath458 .
then @xmath459 it follows from commutativity of diagram and from assumption [ aass : sing ] that the image of the restricted morphism @xmath460 is contained in the divisor @xmath461 . in other words
, there exists a factorisation @xmath462_{b } \ar@/^4mm/[rrrr]^{a|_{v } } & & e \ar[rr]_{\text { , inclusion } } & & x. } \ ] ] equation then follows easily from the composition law for pull - back of torsion - free differentials formulated in lemma [ lem : cltfd ] and from the commutativity of the diagrams considered so far .
@xmath463 this finishes the proof of lemma [ lem : bpx ] .
[ rem : fgh ] the varieties @xmath348 , @xmath464 , @xmath334 and @xmath337 are smooth by assumption [ aass : smooth ] .
the spaces @xmath465 and @xmath466 therefore agree , allowing us to view the torsion - free form @xmath467 as a khler differential , say @xmath468 . since @xmath469 and @xmath470 are morphisms between smooth spaces , the pull - back notions for torsion - free forms and khler differentials agree .
equation therefore implies the following equality of khler differentials @xmath471 we show that @xmath472 can be extended from the open set @xmath337 to all of @xmath334 .
the following criterion , essentially shown in @xcite , will be employed .
[ lem : regcrit ] setup as above .
if there exists a smooth , irreducible variety @xmath473 , a proper , generically finite surjective morphism @xmath474 and a differential form @xmath475 whose restriction to @xmath476 agrees with the pull - back of @xmath472 , @xmath477 then there exists a khler differential @xmath478 that extends @xmath472 . recalling from assumption [ aass : smooth ] that @xmath334 is smooth , the sheaf @xmath344 is locally free , thus reflexive .
we are therefore free to shrink @xmath334 by removing a suitable set of codimension two , and assume without loss of generality that @xmath479 is actually finite .
for finite morphisms between smooth spaces , lemma [ lem : regcrit ] has already been shown in ( * ? ? ?
2.12(ii ) ) . setting as above .
then there exists a khler differential @xmath480 that extends @xmath472 .
recall from remark [ rem : exv ] there always exist smooth varieties @xmath348 and morphisms @xmath386 , @xmath128 forming a diagram as in , where @xmath128 is generically finite . setting @xmath458 as before ,
equation of remark [ rem : fgh ] asserts that the differential form @xmath481 initially defined only on @xmath464 extends to give a regular differential form on all of @xmath348 , namely @xmath482 .
lemma [ lem : regcrit ] therefore applies .
given a smooth variety @xmath348 and morphisms @xmath386 , @xmath128 forming a diagram as in , we need to show that equation holds .
we have seen in lemma [ lem : bpx ] and remark [ rem : fgh ] that the equation holds on @xmath337 . since @xmath334 is smooth by assumption [ aass : smooth ] , the sheaf @xmath344 is torsion - free .
since @xmath337 is open in zariski topology , hence dense , equation holds everywhere , as required .
this finishes the proof of lemma [ lem : const ] . to prove proposition [ prop : const ] , assume that we are given a morphism of klt base spaces , @xmath376 , and a reflexive @xmath89-form on @xmath3 , say @xmath483}_x \bigr)$ ] .
the weak version of proposition [ prop : const ] shown in lemma [ lem : const ] asserts the existence of a resolution map @xmath347 and a reflexive @xmath89-form @xmath378}_z \bigr)$ ] satisfying universal property . to prove proposition [ prop : const ]
, we will show that the form @xmath73 satisfies the stronger requirements of proposition [ prop : const ] . to this end , assume we are given an arbitrary resolution of singularities @xmath484 with form @xmath485 , a smooth variety @xmath486 and morphisms @xmath487 , @xmath488 forming a commutative diagram as in .
we need to show that the following equation holds , @xmath489 to this end , choose a component of the fibred product @xmath490 that surjects onto @xmath3 , and let @xmath491 be a desingularisation of that component .
we obtain a resolution of singularities @xmath492 that dominates both @xmath493 and @xmath493 . in a similar vein , choose a component of @xmath494 that surjects onto @xmath486 .
desingularising , we obtain a smooth variety @xmath348 and a large commutative diagram of morphisms between varieties , @xmath495_{p ' } \ar@{=}[r ] & { \widehat}x \ar[d]_{p } \ar@/^4mm/[dd]^{{\widehat}{}}\\ & & & & & & { \widetilde}x ' \ar[d]_{ ' } & { \widetilde}x \ar[d]_{}\\ v \ar@/^6mm/[rrrrrruu]^{a '' } \ar[rr]_{g''\text { , surjective } } & & v ' \ar@/^3mm/[rrrru]^{a ' } \ar[rr]_{g'\text { , surjective } } & & z_{\operatorname{reg } } \ar[rr]_{f|_{z_{\operatorname{reg } } } } & & x \ar@{=}[r ] & x. } \ ] ] observing that the pull - back of the forms @xmath62 and @xmath496 to the common resolution @xmath491 agree on the preimage of @xmath368 , we obtain an equality of differential forms , @xmath497 an application of lemma [ lem : const ] with @xmath498 , @xmath499 therefore yields the following chain of equalities , @xmath500 equality now follows because @xmath501 is surjective and @xmath502 therefore injective . we have thus shown that @xmath73 satisfies the requirements of proposition [ prop : const ] .
remark [ rem : exv ] asserts the existence of a diagram as in .
choosing one such diagram , it follows from surjectivity of @xmath128 that the restriction - and - pull - back map @xmath503}_z \bigr )
h^0 \bigl ( v,\ , ^p_v \bigr ) , \quad \mapsto { \ensuremath{d_\text{\rm khler}}}g(|_{z_{\operatorname{reg}}})\ ] ] is injective .
equation therefore determines the reflexive form @xmath73 uniquely .
this finishes the proof of proposition [ prop : const ]
assume we are given two functors @xmath504 and @xmath505 that satisfy the conditions of theorem [ thm : pb - thma ] .
given any morphism of klt base spaces , @xmath376 , denote the two associated pull - back maps of reflexive differentials by @xmath506}_x \bigr )
h^0 \bigl ( y,\ , ^{[p]}_y \bigr ) \text{\quad and \quad } { \ensuremath{d_\text{\rm refl}}}^2 f : h^0 \bigl ( x,\ , ^{[p]}_x \bigr ) h^0 \bigl ( y,\ , ^{[p]}_y \bigr),\ ] ] respectively . to prove theorem [ thm : pb - thma ]
, we need to show that @xmath507 this equality will be established in the remainder of the present section [ ssec : pfpbuq ] . before proving equality in general ,
we treat two special cases first .
[ lem : x1 ] if there exists a dense open subset @xmath508 with inclusion @xmath509 such that @xmath510 then equation holds . by functoriality ,
assumption translates as @xmath511 to prove lemma [ lem : x1 ] , it will therefore suffice to show that @xmath512 , and that these maps are injective .
that , however , follows immediately from compatibility with khler differentials , as formulated in diagram of theorem [ thm : pb - thma ] : given any reflexive form @xmath513}_z \bigr)$ ] , then @xmath514 observe that this determines @xmath515 uniquely .
[ lem : x2 ] if @xmath5 is surjective , then equation holds .
it is clear by assumption that the open set @xmath516 is not empty .
compatibility with khler differentials then asserts that @xmath517 , where @xmath518 is the obvious open immersion .
lemma [ lem : x1 ] therefore applies .
next , we treat the general case . to this end , consider a commutative diagram @xmath519^{{\widetilde}f } \ar[d]_{_z } & { \widetilde}x \ar[d]_{}^{\txt{\scriptsize strong resolution of\\\scriptsize singularities } } \\ z \ar[r]_{f } & x } \ ] ] where @xmath520 is a desingularisation of a component of the fibre product @xmath521 that surjects onto @xmath334 .
using lemma [ lem : x1 ] , we may replace @xmath334 with a dense open subset and assume without loss of generality that @xmath334 is smooth .
the following identities are now immediate consequences of compatibility with khler differentials . @xmath522 the equalities have several consequences .
first , we see that @xmath523 since @xmath524 , this immediately implies that @xmath525 but since @xmath526 and @xmath527 are both equal to the standard pull - back of khler differentials , and since @xmath528 is surjective , it is clear that two forms @xmath2 , @xmath73 on @xmath334 agree if and only if @xmath529 . equation therefore implies assertion , finishing the proof of the uniqueness statement in theorem [ thm : pb - thma ] .
given a morphism @xmath330 of klt base spaces , we define an associated pull - back mapping of reflexive differential forms , @xmath340}_x \bigr )
h^0 \bigl ( z,\ , ^{[p]}_z \bigr),\ ] ] by sending a given reflexive form @xmath530}_x \bigr)$ ] to the unique form @xmath531}_z \bigr)$ ] whose existence is asserted in proposition [ prop : const ] .
we need to show that the so - defined @xmath532 satisfies the composition law . to this end , consider a sequence of morphisms between klt base spaces , say @xmath533 and @xmath376 .
we need to show that @xmath534}_x \bigr).\ ] ] to prove , choose desingularisation @xmath535 , choose a component of @xmath536 that surjects onto @xmath520 , and let @xmath537 be a desingularisation of this component .
further , choose a component of @xmath538 that surjects onto @xmath539 , and let @xmath348 be a desingularisation of this component .
in summary , we have constructed smooth spaces @xmath520 , @xmath537 and @xmath348 fitting into a commutative diagram @xmath540^{h } \ar[d]_{q}^{\txt{\scriptsize surjective } } & & { \widetilde}x \ar[dd]_{}^{\txt{\scriptsize resolution of\\\scriptsize singularities}}\\ & & & & { \widetilde}z \ar[d]_{p}^{\txt{\scriptsize resolution of\\\scriptsize singularities } } \\ v \ar[rr]_{g\text { , surjective } } \ar@/^6mm/[rrrruu]^{{\widehat}a } \ar@/^4mm/[rrrru]_a & & z ' \ar[rr]_{f ' } & & z \ar[rr]_f & & x. } \end{split}\ ] ] the following reflexive forms , defined as explained above using proposition [ prop : const ] , will appear in the computation @xmath541 we have seen in remark [ rem : exv2 ] that this definition implies that @xmath62 is the unique differential form on @xmath38 that agrees with the pull - back @xmath2 at points where @xmath381 is isomorphic .
the analogue statement holds for @xmath57 and @xmath73 , so that our notation is consistent with the notation used earlier .
a repeated application of proposition [ prop : const ] , using the fact that two reflexive forms on a normal space agree if they agree on the smooth locus , now shows the following @xmath542 since @xmath128 is surjective , remark [ rem : exv2 ] immediately implies that @xmath361 is injective .
equation therefore follows from the computation . in summary
, we have shown that the definition of pull - back given above does satisfy the composition law .
this finishes the proof of the existence statement in theorem [ thm : pb - thma ] .
in parts [ part:1 ] and [ part:2 ] , we need to discuss torsion sheaves and torsion - free sheaves on reducible spaces .
while no fundamental issues arise , it seems that almost all standard books , such as @xcite , @xcite or @xcite restrict themselves to the irreducible case .
the few existing references touch the subject only very briefly . for completeness sake , we have thus chosen to recall the relevant definitions and to include proofs of all the properties used in this paper . in the setting of notation [ not : ratfct ] ,
the sheaf of rational functions is quasi - coherent .
if @xmath402 is an affine open set , say @xmath544 , then @xmath545 is the ring of rational functions on @xmath3 .
this ring is isomorphic to the localisation @xmath546 , where @xmath547 is the multiplicatively closed set of non - zerodivisors in @xmath548 .
[ defn : torsion ] let @xmath3 be a reduced , quasi - projective scheme and @xmath549 the natural inclusion of the structure sheaf into the sheaf of rational functions . given
a coherent sheaf @xmath86 of @xmath8-modules consider the natural map @xmath550 given as the composition of the following maps , @xmath551 define the _ torsion subsheaf _ of @xmath86 as @xmath552 . the sheaf @xmath86 is called _ torsion sheaf _ if @xmath553 , and _ torsion - free _ if @xmath550 is injective .
[ expl : torgenpts ] in the setting of definition [ defn : torsion ] , let @xmath402 be any affine open set , say @xmath544 . denoting the @xmath548-module associated with the sheaf @xmath86 by @xmath554 , the sheaf @xmath555 of definition [ defn : torsion ] is expressed as follows , @xmath556 in summary , we see that a section @xmath557 is a section of the torsion subsheaf @xmath558 if and only if there exists a non - zerodivisor @xmath559 that annihilates it . in particular
, @xmath2 is a section of the torsion subsheaf if and only if there exists dense open @xmath560 such that @xmath561
. definition [ defn : torsion ] ensures that essentially all properties known from torsion - free sheaves on irreducible spaces also hold in the more general setting of reduced quasi - projective schemes except perhaps the feature mentioned in remark [ rem : tfszoos ] above .
the following properties have been used in this paper .
[ prop : uptf ] let @xmath3 be a reduced quasi - projective scheme , and @xmath562 a morphism of coherent sheaves of @xmath8-modules .
then there exists a unique morphism @xmath563 making the following diagram commutative , @xmath564^{ } \ar[d]_{\text{quotient } } & { \scr{g}}\ar[d]^{\text{quotient } } \\
\factor { \scr{f}}.\operatorname{tor}. \ar[r]_{\widecheck{ } } & \factor { \scr{g}}.{\operatorname{tor}.}. } \ ] ] if @xmath200 is injective , then @xmath563 is injective as well .
[ cor : sstf ] let @xmath3 be a reduced quasi - projective scheme , and @xmath86 a torsion - free coherent sheaf of @xmath8-modules . if @xmath567 is any subsheaf of @xmath8-modules , then @xmath567 is likewise torsion - free .
[ prop : pftf ] let @xmath3 be a reduced quasi - projective scheme . if @xmath568 denotes the inclusion map of one irreducible component , and
if @xmath86 is a torsion - free coherent sheaf on @xmath274 , then @xmath87 is a torsion - free coherent sheaf on @xmath3 .
now assume that @xmath571 , and let @xmath572 be any section , with associated section @xmath573 .
we need to show that @xmath574 , or equivalently @xmath575 . to this end , let @xmath576 denote the union of all irreducible components different from @xmath274 . let @xmath577 be a dense open subset such that @xmath561 .
the set @xmath578 is then open and dense on @xmath274 and @xmath579 . in particular ,
since @xmath86 is torsion - free , this shows that @xmath575 , as claimed .
[ prop : ijtf ] let @xmath3 be a reduced , quasi - projective scheme and @xmath562 a morphism of coherent sheaves of @xmath8-modules . assume that @xmath86 is torsion - free . if @xmath581 is a closed subset such that daniel greb , stefan kebekus , and sndor j. kovcs , _ extension theorems for differential forms , and bogomolov - sommese vanishing on log canonical varieties
_ , compositio math .
* 146 * ( 2010 ) , 193219 , http://dx.doi.org/10.1112/s0010437x09004321[doi:10.1112/s0010437x09004321 ] .
a slightly extended version is available as http://arxiv.org/abs/0808.3647[arxiv:0808.3647 ] .
daniel greb , stefan kebekus , sndor j. kovcs , and thomas peternell , _ differential forms on log canonical spaces _ ,
hautes tudes sci .
publ . math .
* 114 * ( 2011 ) , no .
1 , 87169 , http://dx.doi.org/10.1007/s10240-011-0036-0[doi:10.1007/s10240-011-0036-0 ] an extended version with additional graphics is available as http://arxiv.org/abs/1003.2913[arxiv:1003.2913 ] .
daniel greb , stefan kebekus , and thomas peternell , _ singular spaces with trivial canonical class_. to appear in `` minimal models and extremal rays
proceedings of the conference in honor of shigefumi mori s 60th birthday '' , advanced studies in pure mathematics .
preprint http://arxiv.org/abs/1110.5250[arxiv:1110.5250 ] , october 2011 .
hans grauert and reinhold remmert , _ coherent analytic sheaves _
, grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , vol . 265 , springer - verlag , berlin , 1984 .
mr 755331 ( 86a:32001 ) stefan kebekus , _ differential forms on singular spaces , the minimal model program , and hyperbolicity of moduli stacks _ , in : handbook of moduli , volume ii , editors gavril farkas and ian morrison .
number 26 in advanced lectures in mathematics , international press , beijing , 2013 .
preprint http://arxiv.org/abs/1107.4239[arxiv:1107.4239 ] .
stefan kebekus and sndor j. kovcs , _ the structure of surfaces and threefolds mapping to the moduli stack of canonically polarized varieties _ , duke math
. j. * 155 * ( 2010 ) , no . 1 , 133 , preprint http://arxiv.org/abs/0707.2054[arxiv:0707.2054 ] .
mr 2730371 ( 2011i:14060 )
|
let @xmath0 be a morphism between normal complex varieties , where @xmath1 is kawamata log terminal . given any differential form @xmath2 , defined on the smooth locus of @xmath1
, we construct a `` pull - back form '' on @xmath3 . the pull - back map obtained by this construction is @xmath4-linear , uniquely determined by natural universal properties and exists even in cases where the image of @xmath5 is entirely contained in the singular locus of @xmath1 .
one relevant setting covered by the construction is that where @xmath5 is the inclusion ( or normalisation ) of the singular locus @xmath6 . as an immediate corollary
, we show that differential forms defined on the smooth locus of @xmath1 induce forms on every stratum of the singularity stratification .
the same result also holds for many whitney stratifications .
|
neutron stars are very compact .
their typical masses are @xmath6 ( where @xmath7 is the mass of the sun ) , while their radii are as small as @xmath8 km . as a result ,
a neutron star core contains matter , whose density @xmath9 reaches several @xmath10 ( @xmath11 g @xmath12 being the density of the standard saturated nuclear matter ) .
the core is composed of uniform neutron - rich nuclear matter and extends from @xmath13 to the stellar center ( where @xmath9 can be as high as @xmath14 ) .
it attracts special attention because of its poorly known composition and equation of state ( eos ) ; e.g. , ref . @xcite . from outside , the core is surrounded by a thin ( @xmath15 km thick ) and light ( a few per cent by mass ) crust composed of atomic nuclei , strongly degenerate electrons and ( after the neutron drip at @xmath16 g @xmath12 ) free neutrons . in this paper , we study the shear viscosity of neutron star cores .
it is an important transport property which affects the relaxation of hydrodynamic motions , particularly , a possible differential rotation within the star and stellar oscillations @xcite .
the shear viscosity can be important for damping gravitational wave driven instabilities ( for instance , r - modes ; e.g. , @xcite and references therein ) .
its knowledge is required to analyze the efficiency of such instabilities for generating gravitational waves . for simplicity ,
we consider the cores composed of strongly degenerate neutrons ( n ) , protons ( p ) , electrons ( e ) , and muons ( @xmath17 )
@xmath18-matter , neglecting a possible appearance of hyperons and/or exotic forms of matter ( pion or kaon condensates or quarks or their mixtures ) as predicted by some eoss at @xmath19 ; see , e.g. , ref .
the electrons and muons constitute almost ideal gases .
the muons are absent in the outermost part of the core .
they appear at densities exceeding a threshold value @xmath20 @xcite at which the electron chemical potential reaches the muon rest - mass energy ( @xmath21 ) .
the electrons are ultra - relativistic , while the muons are non - relativistic just after the threshold but become relativistic at higher @xmath9 .
in contrast to electrons and muons , nucleons constitute a strongly interacting fermi liquid where protons are essentially non - relativistic , while neutrons become mildly relativistic at @xmath19 .
the neutrons and protons can be in superfluid state ( e.g. , ref .
@xcite ) . the main contribution to
the shear viscosity @xmath5 in a neutron star core comes from electrons and muons ( lightest and most mobile particles ) and neutrons ( most abundant particles ) , @xmath22 the viscosity @xmath1 of electrons and muons is mainly limited by collisions of electrons and muons between themselves and with other charged particles ( protons , in our case ) via electromagnetic forces .
in contrast , the neutron contribution @xmath2 is limited by neutron - neutron and neutron - proton collisions mediated by strong interactions . as a result , @xmath1 and @xmath2 are nearly independent ( belong to different electromagnetic and nuclear sectors ) and can be calculated separately @xcite . in applications ,
one often employs the viscosity @xmath1 calculated by flowers and itoh @xcite for non - superfluid matter .
recently , andersson et al.@xcite have estimated @xmath1 for superfluid matter . however , these studies neglect an enhancement of collisions of relativistic charged particles due to the exchange of transverse plasmons .
the significance of this effect was demonstrated by heiselberg and pethick @xcite in their study of transport properties of ultra - relativistic quark matter .
recently we ( shternin and yakovlev @xcite hereafter sy07 ) have reconsidered the electron - muon thermal conductivity @xmath23 taking into account the exchange of transverse plasmons .
this effect can reduce @xmath23 by several orders of magnitude . here
we reanalyze @xmath1 in the same manner .
we closely follow sy07 and omit technical details .
in addition , we reconsider @xmath2 , which is a more difficult task involving nucleon - nucleon collisions .
the viscosity @xmath2 was calculated by flowers and itoh @xcite for one eos of non - superfluid matter assuming in - vacuum nucleon - nucleon scattering .
these results were fitted by cutler and lindblom @xcite by a simple analytical expression which is widely used ; according to ref .
@xcite , @xmath24 . recently benhar and valli @xcite
have calculated @xmath2 for pure neutron matter in a self - consistent manner using the same nucleon interaction potential to derive @xmath2 and construct the eos ( also for one eos ) .
we calculate @xmath2 in a more general way than flowers and itoh @xcite .
our approach is similar to that used by baiko , haensel and yakovlev @xcite ( hereafter bhy01 ) for evaluating the thermal conductivity of neutrons .
in addition , we employ recent developments @xcite in calculations of nucleon - nucleon scattering cross sections in nuclear matter . as in bhy01
, we take into account superfluidity of protons .
again , we closely follow the derivation of bhy01 and omit the details . after calculating @xmath1 and @xmath2
, we analyze the shear viscosity in neutron star cores with different eoss .
the shear viscosity is calculated from a system of coupled boltzmann kinetic equations @xmath25 where @xmath26 is the distribution function of momentum - transfer carriers @xmath27 ( with @xmath28 , @xmath17 , or @xmath29 , in our case ) ; @xmath30n , p , e , @xmath17 runs over all particle species ; @xmath31 is the velocity of particles @xmath27 , and @xmath32 is a collision integral , that describes a scattering of particles @xmath27 and @xmath33 : @xmath34 .
\label{collint}\end{aligned}\ ] ] here , 1 and 2 denote particle states before a collision ; 1@xmath35 and 2@xmath35 are particle states after the collision ; @xmath36 is the particle momentum , @xmath37 is the spin state , and @xmath38 is the differential transition probability . the kronecker delta @xmath39 is included to avoid double counting of collisions between identical particles ( @xmath40 ) .
distributions @xmath26 slightly deviate from the equilibrium fermi - dirac distributions @xmath41 owing to the presence of a small hydrodynamical velocity field @xmath42 , @xmath43 where @xmath44 is the particle energy , @xmath45 is its chemical potential , @xmath46 is the temperature , @xmath47 is the boltzmann constant , and @xmath48 measures a deviation from equilibrium .
the electron - muon and neutron transports are decoupled because we neglect electromagnetic interaction between the leptons and neutrons . for calculating @xmath1 ,
the electrons and muons are treated as the only momentum carriers which undergo collisions between themselves and with protons . for calculating @xmath2 , the only momentum carriers are assumed to be neutrons , while the contribution of protons is neglected due to their small fraction .
therefore , the protons are thought to be passive scatterers which obey the equilibrium fermi - dirac distribution .
nonequilibrium parts of the electron , muon , and neutron distributions are found using the standard variational approach with the simplest trial function , @xmath49 where @xmath50 is an effective relaxation time of particles @xmath27 , @xmath31 is their velocity , and @xmath51 with @xmath52 .
the resulting shear viscosity is expressed through the effective relaxation times in a standard way , @xmath53 where @xmath54 , @xmath55 , and @xmath2 are , respectively , the partial electron , muon , and neutron shear viscosities ; @xmath56 is the number density of particles @xmath27 ; @xmath57 is their fermi momentum ; and @xmath58 is an effective mass on their fermi surface .
the electron and muon effective masses differ from their rest masses due to relativistic effects , @xmath59 and @xmath60 .
the neutron and proton effective masses differ from their bare masses mainly due to many - body effects in dense matter ( being determined by neutron and proton densities of state near appropriate fermi surfaces ) . linearizing the kinetic equations , multiplying them by @xmath61 , summing over @xmath62 and integrating over @xmath63
, we obtain a system of equations for the relaxation times , @xmath64 where we introduce the effective collision frequencies , @xmath65 , \label{eq : nuci_int}\\ \nu_{ci}'&=&\frac{3\pi^2\hbar^3}{2 p_{fc}^5 k_bt m_i^*}\int \frac{{\rm d } \bm{p}_1 \,{\rm d } \bm{p}_{1 ' } \,{\rm d } \bm{p}_2\ , { \rm d } \bm{p}_{2'}}{(2\pi\hbar)^{12}}\ , w_{ci}(12|1'2')\ ; f_1f_2(1-f_{1'})(1-f_{2 ' } ) \nonumber\\ & \times & \left[\frac{1}{3}\,p_1 ^ 2p_{2'}^2-\frac{1}{3}\ , p_1 ^
2p_2 ^ 2+(\bm{p}_1 \cdot \bm{p}_2)^2- ( \bm{p}_1\cdot\bm{p}_{2'})^2\right ] , \label{eq : nuci1_int}\end{aligned}\ ] ] with @xmath66 ( the sum is over spin states of all particles 1,2,1@xmath35,2@xmath35 ) .
the formal solution of ( [ eq : coll_syst ] ) for the @xmath18-matter is @xmath67 where @xmath68 in the absence of muons , the expression for @xmath1 simplifies , @xmath69 once collision frequencies are found , the viscosity is obtained from eq .
( [ eta ] ) . in order to determine the collision frequencies from eqs .
( [ eq : nuci_int ] ) and ( [ eq : nuci1_int ] ) one needs to know the transition probability @xmath70 .
the collisions of charged particles should be considered with a proper treatment of plasma screening of electromagnetic interaction .
we discuss the plasma screening and the calculation of @xmath1 in secs .
[ viscosity][corr ] . the neutron viscosity @xmath2 is studied in sec . [ viscn_norm ] . in sec .
[ formalism ] we consider nonsuperfluid nucleons ; the effects of proton superfluidity are analyzed in sec .
[ superflu ] . throughout the paper
we use the simplest variational approach . a comparison with an exact solution
is discussed in sec .
[ s : exact ] .
the plasma screening in neutron star cores is discussed in sy07 . here
, we outline the main points .
the differential collision probability can be written as @xmath71 where @xmath72 is the squared matrix element summed over final and averaged over initial spin states . for collisions of identical particles ,
we have @xmath73 , where the first and second terms correspond to the scattering channels @xmath74 and @xmath75 , respectively . collisions of different particles go through a single channel , @xmath76 , @xmath77 where @xmath78 and @xmath79 are , respectively , momentum and energy transfers in a collision event ; @xmath80 is the transition 4-current ( @xmath81 , 1 , 2 , 3 ) , @xmath82 is the component of @xmath83 transverse to @xmath84 ; @xmath85 is a dirac matrix ; @xmath86 a normalized bispinor ( with @xmath87 ) , and @xmath88 is a dirac conjugate . the first term in eq .
( [ mfi ] ) corresponds to direct coulomb interaction via the longitudinal currents ( with respect to @xmath84 ) ; the space - like longitudinal component of the current is expressed through the time - like component @xmath89 with the aid of charge conservation condition .
the second term describes the interaction via transverse currents .
it is especially important for relativistic particles because @xmath90 .
longitudinal and transverse interactions are accompanied by different plasma screenings described by the functions @xmath91 and @xmath92 in the denominators of eq .
( [ mfi ] ) .
the collision energy and momentum transfers in neutron star cores are typically small , @xmath93 and @xmath94 .
this smallness allows us to use the weak - screening approximation which greatly simplifies the consideration . moreover ,
one typically has @xmath95 , so that it is sufficient to use the asymptotic expressions ( e.g. , sy07 ) @xmath96 where @xmath97 is the fine structure constant ; @xmath98 and @xmath99 are characteristic plasma wavenumbers which depend on plasma composition ( summation is over all types of charged particles ) ; @xmath98 is the familiar thomas - fermi screening wavenumber ; @xmath100 , with @xmath101 in the limit of ultra - relativistic particles .
longitudinal interactions ( via the exchange of longitudinal plasmons ) are mediated by static non - dissipative screening with characteristic wavenumber @xmath98 ( @xmath92 is real ) , while transverse interactions ( via the exchange of transverse plasmons ) are accompanied by the collisionless landau damping ( @xmath91 is purely imaginary ) .
characteristic momentum transfers in transverse interactions are @xmath102 , meaning that such interactions occur on larger spatial scales than the longitudinal ones .
therefore , for relativistic particles , the transverse interactions can be more efficient .
the importance of such interactions was pointed out by heiselberg and pethick @xcite in their study of kinetic properties of relativistic quark plasma .
so far in all calculations of kinetic properties in neutron star cores ( except for sy07 ) the transverse interactions have been erroneously screened by the same static dielectric function @xmath92 as the longitudinal interactions .
this approximation strongly ( up to several orders of magnitude ) overestimates the electron - muon thermal conductivity ( sy07 ) . we will show that it overestimates also @xmath1 ( but less dramatically ) .
the squared matrix element in ( [ eq : wci ] ) for free ultra - relativistic particles can be written as @xmath103 where @xmath104 and @xmath105 are dimensionless functions , @xmath106 @xmath107 being the angle between the vectors @xmath108 and @xmath109 .
the function @xmath105 describes interference between two scattering channels of identical particles . in the weak - screening approximation ,
its contribution is small ; see sec .
[ corr ] .
the collision frequencies are obtained by calculating the integrals ( [ eq : nuci_int ] ) and ( [ eq : nuci1_int ] ) . the calculations are greatly simplified because all particles are strongly degenerate .
it is sufficient to place the colliding particles on their fermi surfaces ( whenever possible ) and use the standard energy - angular decomposition based on @xmath110 , where @xmath111 is the solid angle element in the direction of @xmath36 .
all ( but one ) energy integrations can be done with the aid of the energy - conserving delta - function in ( [ eq : wci ] ) ; only the @xmath112 integration is left .
three angular integrations out of eight are performed with the aid of the momentum - conserving delta - function ; three integrations ( over the position of @xmath113 and over the azimuthal angle of @xmath114 with respect to @xmath113 ) are trivial and give @xmath115 . as a result
, one can reduce the angular integration to the integration over @xmath116 and @xmath117 . then
the collision frequencies ( [ eq : nuci_int ] ) and ( [ eq : nuci1_int ] ) can be written as @xmath118 ^ 2 } i_{\omega ci}(\omega ) , \\
\nu'_{ci}&=&\frac{12 \hbar^2\alpha^2}{\pi^2 p_{fc}^5 m_i^ * c^2 } ( k_b t)^2\int_0^\infty { \rm d } w\frac{w^2 \exp(-w)}{\left[1-\exp(-w)\right]^2 } i'_{\omega ci}(\omega),\end{aligned}\ ] ] where @xmath119 .
the functions @xmath120 and @xmath121 are the angular integrals @xmath122 where @xmath123 is the maximum momentum transfer in a collision event .
owing to a trivial integration over @xmath107 , @xmath124 contains two terms coming from @xmath125 and @xmath126 , while @xmath127 contains only the contribution from @xmath128 , @xmath129 let us calculate the angular integrals in the leading approximation with respect to the parameters @xmath130 and @xmath131 .
this approximation is always justified for the transverse interactions because of the presence of a small quantity @xmath132 in the expression for @xmath130 .
however , it is less accurate for the longitudinal contribution since @xmath131 is not too small ; we will discuss corresponding corrections in sec .
[ corr ] .
the leading - order expressions for the angular integrals are @xmath133 note that the leading - order expression for @xmath134 is independent of @xmath135 , being of the same order of magnitude with respect to @xmath131 as @xmath136 .
in contrast , @xmath137 .
the final integration over @xmath135 gives the collision frequencies @xmath138 where @xmath139 , @xmath140 is the riemann zeta function , and @xmath141 is the gamma function .
equations ( [ eta ] ) , ( [ taus])([nomuons ] ) , and ( [ nuper])([nucrossed ] ) give @xmath1 in the weak - screening approximation . for typical conditions in a neutron star core , @xmath142 however , the inequality is not so strong as for the thermal conductivity ( sy07 ) .
the dominance of @xmath143 over @xmath144 is determined by the factor @xmath145^{1/3}$ ] which increases slowly with decreasing @xmath46 .
it is more accurate to include all components of the collision frequencies . for the thermal conductivity problem
, we had @xmath146 , so that transverse interactions dominated at all temperatures of interest ( sy07 ) . nevertheless , for the not too high temperatures ( see sec.[results ] for details )
, @xmath1 is mainly determined by the collisions via the exchange of transverse plasmons . in this case , the electron and muon momentum transports are decoupled [ see eq .
( [ taus ] ) ] , @xmath147 then the shear viscosity of electrons or muons ( @xmath148 e or @xmath17 ) becomes @xmath149 we see , that in the low - temperature limit , @xmath1 has a non - standard temperature behavior , @xmath150 ( instead of the standard fermi - liquid dependence @xmath151 ) .
the non - standard behavior was pointed out by heiselberg and pethick @xcite for an ultra - relativistic quark plasma .
our results involve collisions of charged particles in the @xmath18-matter ( for any degree of relativity of muons ) .
our expressions for @xmath1 depend only on the number densities of charged particles and on their effective masses ; therefore , they can be used for any eos of dense matter .
previous calculations @xcite overestimated @xmath152 because they employed the improper plasma screening of transverse interactions . equation ( [ eq : etaperp ] )
remains valid in the presence of other charged particles ( such as @xmath153 hyperons ) .
as mentioned above , the corrections to @xmath154 containing higher - order powers of @xmath130 can be neglected , so that @xmath154 can be taken in the form ( [ tauperp ] ) .
in contrast , the corrections to @xmath155 containing higher - order powers of @xmath131 can be important . at not too small temperatures , at which @xmath155 can give a noticeable contribution , such corrections can affect @xmath1 .
we will discuss several corrections of this type .
the main corrections to the leading terms arise from the @xmath158-dependence of the functions @xmath104 [ eqs.([phiper ] ) and ( [ phiparper ] ) ] and from the @xmath158-dependence in eqs .
( [ eq : intci ] ) and ( [ intci1 ] ) .
the integral @xmath159 is calculated precisely , @xmath160i^\parallel_4(q_m / q_l)\nonumber \\ & + & \frac{\pi q_l^3}{16 } \left[(m_c^{*2}+m_i^{*2})c^2+p_{fc}^2\right ] i^\parallel_6(q_m / q_l)-\frac{\pi\hbar^2 q_l^5}{64 } i^\parallel_8(q_m / q_l ) , \end{aligned}\ ] ] where we have introduced the integrals @xmath161 whose expressions are given in appendix [ appendix ] .
after the energy integration the corrected collision frequency becomes @xmath162 similar corrections should be calculated for @xmath157 . in the leading - order approximation
( with respect to @xmath130 ) , @xmath163 can be neglected in the denominator of @xmath128 .
the remaining angular integral is taken , @xmath164,\ ] ] where @xmath165 after the energy integration we finally obtain @xmath166 our calculations show that these corrections to @xmath156 and @xmath157 can reach @xmath16770% .
it is advisable to include them in @xmath1 .
so far we have considered the function @xmath104 calculated for free relativistic particles .
it is a good approximation for collisions within the electron - muon subsystem , because the electrons and muons constitute almost ideal fermi gases .
however , the protons belong to a strongly interacting fermi liquid ; this case should be analyzed separately .
first of all we notice that the protons are non - relativistic .
moreover , we will assume , that many - body effects can be treated by introducing an effective proton mass @xmath168 .
under these assumptions , the proton transition current can be written as @xmath169 , which only slightly modifies @xmath104 . the expression ( [ phiparper ] ) for @xmath128 remains the same , while the two other functions become @xmath170 the difference between @xmath171 , calculated with ( [ phicppar ] ) and ( [ phipar ] ) , is proportional to some power of @xmath172 .
contrary to @xmath131 , this ratio is always small for the conditions in neutron star cores .
hence eq .
( [ nuparfull ] ) with the angular integral ( [ intparfull ] ) remains a valid approximation .
for the completeness of our analysis , we present the modified angular integral , @xmath173 this expression gives almost the same @xmath174 .
the last correction to be discussed is the correction to @xmath175 ( to @xmath176 and @xmath177 ) due to the interference between two scattering channels ( @xmath178 and @xmath179 ) for collisions of identical particles .
this interference is described by the dimensionless function @xmath105 which ( like @xmath104 ) contains longitudinal , transverse and mixed components . an accurate consideration shows that all these components are smaller than corresponding components of @xmath104 .
therefore , the interference correction to @xmath180 can be neglected , and noticeable corrections can arise only to @xmath181 and @xmath182 .
these corrections have been calculated in the same way as in previous sections .
we have obtained that they are numerically small ( give @xmath183% contribution to @xmath184 ) .
their contribution to @xmath1 is always negligible as expected without any numerical calculations .
such corrections can be significant under two conditions .
first , the longitudinal component @xmath181 should be comparable to @xmath180 .
second , @xmath175 itself should give a noticeable contribution to @xmath1 .
the former condition would be realized at high temperatures if particles @xmath27 are weakly - relativistic .
the electrons are ultra - relativistic in neutron star cores and do not obey the above requirement .
the muons can be weakly - relativistic there , but if they are their contribution to @xmath1 is not large .
the importance of the interference corrections in @xmath185 collisions is further reduced by a ( typically ) stronger contribution of @xmath27p collisions .
there are also collisions between electrons and muons .
the interference corrections for such collisions are absent ; corresponding partial collision frequencies are of the same order of magnitude as @xmath175 .
thus , the corrections to @xmath186 and @xmath175 seem to be negligible .
the kinematical corrections to @xmath156 and @xmath157 are significant if @xmath144 can not be neglected in comparison with @xmath143 . note that in sy07 , for the electron - muon thermal conductivity problem , no corrections have been required because the thermal - conduction frequencies @xmath143 dominate at any density and temperature of practical interest .
so far we have used a simplest variational solution for the shear viscosity based on the expression for the trial function ( [ phi ] ) with @xmath187 independent of the particle energy @xmath44 . actually , however , the energy dependence of @xmath48 is more complicated , which affects the shear viscosity .
it is convenient to introduce a correction factor @xmath188 that relates the exact and variational shear viscosities , @xmath189 in ordinary fermi - systems , where the collision probability is independent of energy transfer @xmath190 , the factor @xmath188 can be calculated using the theory developed by sykes and brooker @xcite for one component systems and extended by flowers and itoh @xcite and anderson et al .
@xcite for multicomponent systems .
unfortunately , this theory can not be directly applied to our case because of the dynamical character of transverse plasma screening ( landau damping ) .
the factor @xmath188 for the thermal conductivity with account for the exchange of transverse plasmons was estimated in sy07 .
let us do the same for the shear viscosity .
as in sy07 , we restrict ourselves to the exchange of transverse plasmons in the weak - screening approximation . then the electron transport decouples from the muon one , and we can consider one type of momentum carriers .
we redefine @xmath48 in ( [ phi ] ) as @xmath191 where @xmath192 is an effective relaxation time ( that can be treated as a normalization constant ) , and an unknown function @xmath193 of @xmath194 describes the energy dependence of @xmath48 . substituting ( [ phi_exact ] ) into the linearized kinetic equation ,
one obtains an integral equation for @xmath193 , @xmath195 + i_{\omega c}(x'-x ) \psi(x')\right\ } \label{inteq_exact},\end{aligned}\ ] ] where @xmath196 , @xmath197 , and @xmath198 the integral equation ( [ inteq_exact ] ) is more complicated than that for the thermal conductivity ( see eq . ( 42 ) in sy07 ) .
the term with @xmath199 in ( [ inteq_exact ] ) appears because we go beyond the simplest variational approach of @xmath200 . without that approach ,
there is no cancellation of zero - order expansion terms ( in series of @xmath158 ) in kinematical factors in eq.([eq : nuci_int ] ) .
it was that cancellation which led to the appearance of the @xmath201 term under the integral in eq .
( [ eq : intci ] ) . the integral @xmath202 coincides ( save constant factor ) with the angular integral for the thermal conductivity problem ( eq . ( 25 ) in sy07 ) . taking the weak - screening expressions for @xmath203 and @xmath199 with the exchange of transverse plasmons alone , and choosing @xmath204
we obtain the dimensionless equation @xmath205 , \label{eqpsi_exact}\end{aligned}\ ] ] where @xmath206 .
the quantity @xmath207 is the transverse screening wavenumber @xmath208 , with @xmath209 replaced by @xmath210 . in a neutron star core ,
one typically has @xmath211 , and hence @xmath212 . if @xmath213 , then the weak - screening approximation is not justified .
once @xmath193 is found by solving eq .
( [ eqpsi_exact ] ) , the shear viscosity is given by @xmath214 we have solved eq .
( [ eqpsi_exact ] ) numerically and compared the result with the variational one ( @xmath215 $ ] ) . for @xmath216 , we obtain @xmath217 . for @xmath218 , the factor @xmath219 becomes nearly independent of @xmath220 .
therefore , we have @xmath221 indicating that the simplest variational approach is well justified . were the electron ( and muon ) collisions determined solely by the exchange of longitudinal plasmons ( with the transition matrix element independent of @xmath112 ) , one could find @xmath188 from the standard theory . in that case one also obtains @xmath221 ( see , e.g. , ref .
we expect , that in the most general case , when electron and muon collisions are governed by the exchange of transverse and longitudinal plasmons , the correction factor @xmath188 differs from @xmath222 by @xmath223 . if so , the simplest variational approach is sufficiently accurate , and no corrections are required ( @xmath222 ) for the majority of astrophysical applications . in this section
we calculate the neutron shear viscosity @xmath2 .
we employ the same formalism as was used in bhy01 for studying the neutron thermal conductivity .
similar approach was used by baiko and haensel @xcite to determine kinetic coefficients mediated by neutron - neutron collisions .
we calculate @xmath224 from eq .
( [ eta ] ) ; the effective relaxation time of neutrons , @xmath225 , is given by eq .
( [ taus ] ) , being determined by the collision frequencies ( [ nus ] ) of neutrons with neutrons and protons .
the neutron - neutron collision frequency can be written as @xmath226 ( note that the authors of bhy01 did not separate @xmath227 and @xmath228 but considered their sum ) .
the neutron - proton collision frequency is @xmath229 here , @xmath230 is a bare nucleon mass . the quantities
@xmath231 and @xmath232 are the effective nucleon - nucleon scattering cross - sections introduced in eq .
( 22 ) of bhy01 ( for the thermal conduction problem ) . for the shear viscosity , in the same notations as in bhy01
, we obtain @xmath233 where @xmath234 , @xmath235 , @xmath236 , and @xmath237 .
this choice of integration variables is convenient for numerical integration .
the quantities @xmath238 and @xmath239 are squared matrix elements for nucleon - nucleon scattering ( in the notations of bhy01 , @xmath240 and @xmath241 ) .
let us emphasize that kinematic restrictions in eqs .
( [ e : snn ] ) and ( [ e : snp ] ) are very different .
the effective cross section @xmath231 is determined by a wide spectrum of momentum transfers @xmath158 ( or , equivalently , of scattering angles ) .
our calculations show that one can get a reasonably accurate @xmath231 assuming that @xmath238 is independent of @xmath158 .
such a @xmath158-averaged @xmath238 can be extracted from a total neutron - neutron scattering cross section .
in contrast ( because , typically , @xmath242 ) , @xmath232 is determined by small momentum transfers , that is by a small - angle cross section of neutron - proton scattering . by analogy with bhy01 , we can write @xmath243 and @xmath244 . here , @xmath245 and @xmath246 are the effective cross sections calculated with in - medium squared matrix elements , @xmath247 and @xmath239 ; @xmath248 and @xmath249 are similar cross sections calculated with the in - vacuum matrix elements ; @xmath250 and @xmath251 are the ratios of the in - medium to the in - vacuum cross sections .
the authors of bhy01 calculated all these quantities for the thermal conduction problem .
the squared matrix elements @xmath238 and @xmath239 were extracted from nucleon - nucleon differential scattering cross sections calculated in refs .
@xcite for symmetric nuclear matter with the bonn nucleon - nucleon interaction potential using the dirac - brueckner approach .
an accurate extraction of the in - medium @xmath238 and @xmath239 required the knowledge of effective masses @xmath252 and @xmath253 ( not reported in @xcite ) .
for that reason , the procedure used in bhy01 was ambiguous .
thus , the factors @xmath250 and @xmath251 , presented in bhy01 for the neutron thermal conductivity , are model dependent and not very certain .
now we turn to calculating @xmath245 , @xmath246 , @xmath254 , @xmath249 , @xmath250 , and @xmath251 for the shear viscosity . to avoid the above drawbacks
, we suggest to neglect the in - medium effects on the squared matrix elements and set @xmath255 , @xmath256 , and @xmath257 .
according to bhy01 ( for the thermal conductivity ) , @xmath250 and @xmath251 are indeed @xmath15 ( and @xmath251 is relatively unimportant ) .
recently zhang et al .
@xcite have studied nucleon - nucleon scattering cross sections in nuclear matter taking into account two - nucleon and three - nucleon interactions .
they used the brueckner - hartree - fock approach and the argonne v14 nucleon - nucleon interaction model supplemented by three - nucleon interactions .
their principal conclusion is that the in - medium effects on square matrix elements are relatively weak , while the main medium effect consists in modifying ( mostly reducing ) @xmath258 and @xmath253 .
the reduction of effective masses under the simultaneous effects of two - nucleon and three - nucleon forces is much stronger than under the effect of two - nucleon forces alone .
thus , we have calculated @xmath259 and @xmath260 from eqs .
( [ e : snn ] ) and ( [ e : snp ] ) using the in - vacuum matrix elements @xmath238 and @xmath261 from refs . @xcite
. these matrix elements accurately reproduce @xcite well elaborated laboratory measurements of differential nucleon - nucleon scattering cross sections .
our calculations of @xmath259 and @xmath262 are expected to be very close to those done with in - vacuum cross sections measured in laboratory . in this sense ,
our values of @xmath259 and @xmath260 are model independent .
similar calculations of @xmath259 in ref .
@xcite give slightly different results due to the different data sets for @xmath247 .
because our equations ( [ e : nunnfinal ] ) and ( [ e : nunp ] ) for the nucleon - nucleon collision frequencies contain a proper dependence on @xmath252 and @xmath253 , they can be regarded as _ independent _ of any specific model for nucleon - nucleon interaction .
it can be a two - body or two - body plus three - body interaction ; its explicit form is not essential .
thus , we obtain a description of the neutron shear viscosity valid for any eos of nucleon matter .
this approach is not strict ( uses the in - vacuum matrix elements ) but universal .
one can in principle calculate more accurate in - medium matrix elements for any chosen eos but loosing the universality . for calculating the diffusive thermal conductivity from the equations of bhy01
, we would recommend to adopt the same approach and set @xmath255 in those equations .
the results of our calculations can be fitted by the expressions @xmath263 where @xmath264 is the fermi wavenumber of nucleons @xmath33 expressed in @xmath265 . as in bhy01
, the calculations and fits cover the range of @xmath266 from 1.1 to 2.6 @xmath265 and the range of @xmath267 from 0.3 to 1.2 @xmath265 .
these parameter ranges are appropriate to neutron star cores at @xmath268 .
the fit errors for @xmath231 do not exceed @xmath269 .
the maximum fit error of @xmath232 is @xmath270 ( at @xmath271 @xmath265 and @xmath272 @xmath265 ) .
neutrons and protons in the @xmath18-matter of neutron star cores can be in superfluid state ( e.g. , ref .
@xcite ) . here , we study the effects of superfluidity on the shear viscosity .
let @xmath273 be the critical temperature for superfluidity of neutrons , and @xmath274 be the same for protons .
proton superfluidity means superconductivity .
calculations of superfluid critical temperatures are complicated and very sensitive to a chosen model of nucleon - nucleon interaction and a method to employ many - body ( polarization ) effects @xcite .
numerous calculations give drastically different @xmath275 and @xmath276 .
it is instructive not to rely on any particular model but treat @xmath277 and @xmath278 as free parameters varied within reasonable limits ( @xmath279 k ) , in accordance with microscopic calculations .
neutron superfluidity has no direct effect on the shear viscosity @xmath1 of electrons and muons ( because @xmath1 is limited by electromagnetic interactions ) .
however , it strongly affects neutron star hydrodynamics in a complicated way .
it makes the hydrodynamics essentially multifluid ( with several hydrodynamical velocity fields ; e.g. , @xcite and references therein ) ; it introduces an entrainment effect ( which relates motion of neutrons and protons ) , creates a very specific spectrum of elementary medium excitations ( phonons ) and associated specific energy and momentum transfer mechanisms .
all these problems go far beyond the scope of our paper .
therefore , we will neglect the effects of neutron superfluidity ( will treat neutrons as normal ) but consider the effects of proton superfluidity on @xmath4 ( assuming the protons to be passive scatterers of neutrons , i.e. , ignoring momentum transport by protons ) .
we disregard thus hydrodynamical effects of neutron and proton superfluids .
microscopically , proton superfluidity manifests itself in rearranging proton states ( from normal fermi - liquid quasiparticles to bogoliubov quasiparticles ) and in the appearance of a gap @xmath280 in the proton energy spectrum near the fermi level ( @xmath281 ) , @xmath282 where @xmath283 ; the presented equation is valid at @xmath284 .
it is generally believed , that cooper paring of protons appears in the singlet @xmath285s@xmath286 state ( e.g. , ref .
@xcite ) . the temperature dependence of @xmath280 , calculated in the bcs approximation , can be approximated as @xcite @xmath287 where @xmath288 .
the effects of proton superfluidity on @xmath1 are twofold .
first , proton superfluidity affects the plasma dielectric function and , hence , the screening of electromagnetic interactions .
the longitudinal dielectric function is almost insensitive to the presence of superfluidity @xcite , while the transverse dielectric function modifies collision frequencies @xmath143 .
the frequencies @xmath289 remain almost unchanged because they are independent of the transverse screening in the leading order . as in sy07 , a collision frequency in superfluid matter ( to be denoted as @xmath290 ) can be written as @xmath291 where @xmath143 ( @xmath30e , @xmath17 ) stands for a collision frequency in non - superfluid matter , while @xmath292 accounts for the superfluid effects ( which mainly reduce the collision rate ) ; @xmath293 is a slowly varying function determined by plasma composition .
we have @xmath294 in the absence of muons , and @xmath295 in the presence of muons , with the maximum value of @xmath296 in the limit of ultra - relativistic muons . the reduction factor
@xmath297 for the shear viscosity is , however , not the same as for the thermal conductivity ( obtained in sy07 ) and will be calculated below . the second effect of proton superfluidity consists in an additional ( direct ; not through plasma screening ) reduction of the lepton - proton collision frequencies ( such as @xmath298 and @xmath299 ) .
this direct reduction is exponential ; it can be described by the reduction factors @xmath300 the reduction factors @xmath301 and @xmath302 are not the same due to the difference in longitudinal and transverse plasma screenings . following sy07 , it is convenient to introduce the reduction factor @xmath303 for the total transverse collision frequency @xmath304 , @xmath305/(r+1).\end{aligned}\ ] ] below we calculate @xmath306 , @xmath307 , @xmath308 , and @xmath303 .
superfluid reduction of lepton - lepton collisions is governed by the transverse polarization function @xmath91 .
for the conditions in neutron star cores , it is sufficient to use @xmath91 in the so - called pippard limit ( @xmath309 , @xmath310 and @xmath311 , where @xmath312 is the coherence length ) . in this approximation , the proton contribution to @xmath91 reads @xmath313 where @xmath314 , and @xmath315 is the response function calculated in ref .
@xcite and discussed in sy07 in more details . in the non - superfluid limit of @xmath316 one has @xmath317 , which corresponds to the standard landau - damping expression . in the opposite case of strong superfluidity ( @xmath318 ) ,
the response function @xmath315 becomes pure real , @xmath319 . for intermediate superfluidity , @xmath320 ,
we have used the expressions for @xmath315 derived in @xcite .
they are valid for a pure bcs formalism neglecting collective modes and related vortex renormalization in current operators due to gradient invariance . however , as in sy07 , the main contribution to @xmath1 comes from the parameter values far from characteristic frequencies of collective modes ( far from @xmath321 ) and we can use the standard bcs theory .
the expression for the total polarization function in the superfluid case takes the form @xmath322 \right\}.\ ] ] in the case of strong superfluidity , the main contribution to @xmath323 comes from protons . moreover , the character of plasma screening changes . instead of
the dynamical landau damping , the screening becomes static , with the frequency - independent screening wave number @xmath324^{1/3}$ ] . in neutron star cores ,
one typically has @xmath325 .
therefore , the relation @xmath326 remains true in the superfluid case .
in other words , the exchange of transverse plasmons in proton superfluid remains more efficient than the exchange of longitudinal plasmons .
the strong inequality @xmath327 justifies the use of the leading - order weak screening approximation in describing the exchange of transverse plasmons . in order to calculate the reduction factor @xmath306
one should reconsider the transverse angular integral @xmath328 taking into account the changes of electrodynamical plasma properties in superfluid matter . in the leading order with respect to @xmath329 , @xmath330 where @xmath331 refers to non - superfluid matter , while @xmath332^{1/3 } \left[\left(\pi w r+y\im q(w , y)\right)^2 + \left(y\re q(w
, y)\right)^2\right]^{1/3}}{|\pi w r+\delta\im q(w , y)| } \nonumber\\ & \times&\frac{2}{\sqrt{3 } } \sin\left[\frac{2}{3}\arctan\frac{|\pi w r+y\im q(w , y)|}{y\re\ , q(w , y ) } \right]\end{aligned}\ ] ] accounts for superfluid effects . in the limit of strong superfluidity ( @xmath333 )
we have @xmath334^{1/3}.\end{aligned}\ ] ] this asymptotic @xmath135-dependence compensates the @xmath135-dependence in @xmath335 ( that appeared under the effect of plasma screening ) . moreover , in the expression for @xmath336 the collision energy @xmath190 is now replaced by the energy gap @xmath280 .
finally , we write @xmath337 , and the reduction factor becomes @xmath338 ^ 2 } \ ,
w^{5/3 } f^{\perp}(w , y , r)\ , { \rm d}w.\end{aligned}\ ] ] when superfluidity vanishes ( @xmath339 ) we evidently have @xmath340 .
in the opposite case of strong superfluidity ( @xmath333 ) we obtain @xmath341 thus , strong proton superfluidity restores the temperature dependence @xmath342 that is standard for fermi systems .
this result was derived in sy07 for the thermal conduction problem .
it is a natural consequence of changing plasma screening from dynamical to statical one when @xmath46 falls below @xmath278 . in addition , we have computed @xmath292 for a wide grid of @xmath343 .
we do not present an appropriate fit , because we will calculate and fit the total reduction factor @xmath344 for @xmath154 .
now consider a direct effect of superfluidity on electron - proton and muon - proton collision rates .
the consideration is similar to that for the thermal conduction problem ( sy07 ) .
the proton energy gap has to be included in the expressions for the collision frequencies through the proton fermi - dirac distributions .
in addition , the electron - proton and muon - proton scattering matrix elements have to be calculated using wave functions of proton bogoliubov quasiparticles . as a result
, the reduction factors @xmath345 and @xmath346 can be written as @xmath347[\exp(z_{2'}-z_2)-1 ] } f^\perp(|z_{2'}-z_2|,y , r)\right.\nonumber\\ & -&\left .
\frac{(z_{2'}+z_2)|z_{2'}+z_2|^{-1/3 } \left(1 - 4 \rm{u}_2\rm{u}_{2'}\rm{v}_2 \rm{v}_{2'}\right ) } { [ 1+\exp(z_{2'})][\exp(-z_{2'}-z_2)-1 ] } f^\perp(|z_{2'}+z_2|,y , r ) \right\},\\ r^\parallel_{\rm p}(y)&= & \frac{3}{\pi^2}\int\limits_0^\infty\int\limits_0^\infty \frac{{\rm d}x_2\,{\rm d}x_{2'}}{1+\exp(z_2 ) } \left\{\frac{(z_{2'}-z_2)(1 - 4\rm{u}_2\rm{u}_{2'}\rm{v}_2\rm{v}_{2 ' } ) } { [ 1+\exp(-z_{2'})][\exp(z_{2'}-z_2)-1 ] } \right.\nonumber\\ & -&\left .
\frac{(z_{2'}+z_2)(1 + 4\rm{u}_2\rm{u}_{2'}{\rm v}_2{\rm v}_{2 ' } ) } { [ 1+\exp(z_{2'})][\exp(-z_{2'}-z_2)-1 ] } \right\},\end{aligned}\ ] ] where @xmath348 @xmath349 and @xmath350 . in the limit of strong superfluidity ( @xmath333 )
we obtain @xmath351 and @xmath352 , where @xmath353 and @xmath354 thus , at @xmath355 collisions with superfluid protons are exponentially suppressed
. then the shear viscosity @xmath356 is limited by collisions within the electron - muon subsystem ( which are also affected by proton superfluidity as described in sec .
[ nu_emu_superfl ] ) .
we have computed @xmath357 for a wide range of @xmath343 and fitted the results by the expression @xmath358\right\ } \nonumber\\ & \times & \exp\left[-\sqrt{(0.22)^2+y^2}\right ] , \label{e : rparsfit}\end{aligned}\ ] ] which reproduces also the asymptotic limits .
the maximum relative fit error is @xmath359 at @xmath360 .
we do not present a separate fit expression for @xmath361 , but give the fit of the total reduction factor @xmath362 : @xmath363 , \\
g_1&=&0.87 - 0.314r,\ \ \ \ g_2=(0.423 + 0.003r)y^{1/3}+0.0146y^2 - 0.598y^{1/3}\exp(-y ) , \nonumber \\
g_3&=&251 r^{-9}(r+1)^6(1-g_1)^9 , \nonumber\end{aligned}\ ] ] with the maximum fit error @xmath364 at @xmath294 and @xmath365 .
this fit reproduces also the limiting case of @xmath366 at @xmath367 , and the asymptote at @xmath318 , @xmath368 recently the electron shear viscosity in superfluid matter has been analyzed by andersson et al .
these authors have used the standard ( but approximate ) approach in which the transverse plasma screening is assumed to be the same as the longitudinal one .
this approach is inaccurate even in the non - superfluid case . in ref.@xcite
the effects of superfluidity are described by a reduction factor @xmath369 for the effective electron - proton collision frequency .
that factor has been taken from ref .
@xcite devoted to the thermal conductivity problem . however , the reduction factors for the thermal conductivity and shear viscosity are different .
moreover , the factor @xmath369 in @xcite is inaccurate even for the thermal conductivity , because it assumes approximate plasma screening and neglects additional terms associated with creation / annihilation of proton bogoliubov quasiparticles ; see sy07 for details .
nevertheless , numerical values of @xmath54 , derived from the results of @xcite for superfluid matter , are not too different from our results .
typically , they overestimate @xmath54 by a factor of three , and this overestimation increases with decreasing @xmath370 . in the limit of strong superfluidity , the temperature dependence of @xmath371 formally restores the standard fermi - liquid behavior , @xmath372 . therefore , @xmath156 can be comparable to @xmath371 .
note , that the ratio @xmath373 in superfluid matter remains approximately the same as its value at @xmath374 .
if @xmath370 is sufficiently small , then at @xmath375 we have @xmath376 , and the same inequality holds at smaller @xmath46 .
the shear viscosity in superfluid matter , fully determined by the exchange of transverse plasmons , can be written as @xmath377 one can use this expression for estimates , but we recommend to employ the total collision frequency in practical calculations .
we study the effect of proton superfluidity on neutron - proton collisions .
even this problem is difficult and we adopt a simplified approach used in bhy01 for the problem of neutron thermal conductivity .
it has also been widely used for analyzing superfluid suppression of various neutrino processes ( e.g. , @xcite and references therein ) .
it consists in taking an ordinary differential probability of a given scattering process ( neutron - proton scattering , in our case ) and inserting particle energies ( [ e : disperssf ] ) with energy gaps in corresponding fermi - dirac distribution functions . in our case
, this approach is expected to be sufficiently accurate .
let us recall that we consider the protons only as neutron scatterers .
proton superfluidity suppresses this scattering channel , and our approach reproduces such a suppression . in this approximation , the neutron - proton collision frequency becomes @xmath378 where @xmath379 is the superfluid reduction factor .
the latter factor is given by the same expression as the reduction factor for lepton - proton collisions , @xmath380 , save the coherence factors , @xmath381[\exp(z_{2'}-z_2)-1 ] } \right.\nonumber\\ & -&\left .
\frac{z_{2'}+z_2 } { [ 1+\exp(z_{2'})][\exp(-z_{2'}-z_2)-1 ] } \right\}.\end{aligned}\ ] ] it obeys the asymptotes @xmath382 and @xmath383 at @xmath384 , where @xmath385 .
we have calculated @xmath379 in a wide range of @xmath343 and fitted the results by the expression @xmath386 \exp \left [ 2.26-\sqrt{(2.26)^2+y^2 } \right ] \nonumber\\ & + & \frac{1}{3}\left(1 + 0.00056\,y^4\right ) \exp \left [ 6.2-\sqrt{(6.2)^2 + 4\,y^2 } \right ] ; \end{aligned}\ ] ] the formal maximum fit error is @xmath387 at @xmath388 . note , that proton superfluidity affects @xmath2 weaker than @xmath1 ( because of a relatively small contribution of neutron - proton collisions to @xmath2 ) .
our results can be used for a wide range of eoss of the npe@xmath17-matter in neutron star cores . for illustration , we have selected five model eoss .
the parameters of these eoss are given in table [ tab : eos ] , including the maximum gravitational mass @xmath389 of stable stars and the threshold density @xmath390 of muon appearance .
.parameters of the selected eoss : the compression modulus @xmath391 of symmetric saturated nuclear matter ; the muon threshold density @xmath390 ; and also the central density @xmath392 , the mass @xmath393 and radius @xmath394 of maximum - mass models ( @xmath390 and @xmath392 are given in units of @xmath395 g @xmath12 ) [ cols="^,^,^,^,^,^ " , ] the apr eos was constructed by akmal , pandharipande , and ravenhall @xcite ( their model argonne v18+@xmath396+uix@xmath397 ) ; it is often used in the literature .
specifically , we adopt its convenient parametrization proposed by heiselberg and hjorth - jensen @xcite and described as apr i by gusakov et al . @xcite . it is sufficiently stiff ,
the maximum neutron star mass is @xmath398 ( and the maximum - mass star has circumferential radius of @xmath399 km ) , the muons appear at @xmath400 g @xmath12 ; see table [ tab : eos ] .
the pal eoss are convenient semi - analytical phenomenological eoss proposed by prakash , ainsworth , and lattimer @xcite .
they differ by the functional form of the dependence of the symmetry energy @xmath401 of dense matter on the baryon number density @xmath402 .
this dependence is described @xcite by a function @xmath403 , where @xmath404 , @xmath405 @xmath406 being the baryon number density of saturated symmetric matter . for the pal eoss
i , ii , and iii , these functions are @xmath407 , @xmath408 , and @xmath409 , respectively .
the pal iv eos belongs to the same family of eoss , but with the symmetry energy @xmath410 suggested by page and applegate @xcite .
the pal eoss differ also @xcite by the value of the compression modulus @xmath391 of saturated symmetric matter , @xmath411 , @xmath412 , and @xmath413 mev .
nevertheless , the particle fractions @xmath414 as a function of @xmath402 are independent of @xmath391 ( for these eoss ) ; the dependence of @xmath402 on @xmath9 is almost identical for the three selected @xmath391 values [ at a fixed @xmath403 ] .
hence , the collision frequencies and the shear viscosity are independent of @xmath391 . however , taking different @xmath391 , one obtains very different neutron star models ( different mass - radius relations and @xmath393 ; see table [ tab : eos ] ) . for illustration ,
we take @xmath416 mev for all pal models ( unless the contrary is indicated ) . therefore ,
our selected eoss correspond to a large variety of neutron star models . of electrons and muons versus density @xmath417 ( in units of @xmath395 g @xmath12 ) for different eoss ( table [ tab : eos ] ) at @xmath418 k ( @xmath419 ) .
the thin solid line ( apr-@xmath420 ) shows the viscosity @xmath1 calculated with account for the exchange of longitudinal plasmons alone .
, scaledwidth=50.0% ] figure [ fig : viscem ] shows the shear viscosity @xmath1 of electrons and muons versus density at @xmath418 k for five eoss .
the given temperature is typical for middle - aged ( @xmath421 yr ) isolated ( cooling ) neutron stars without enhanced neutrino emission in their cores ( e.g. , refs .
the proton effective mass is taken to be @xmath419 .
the thick lines give @xmath1 for the apr and pal i iv eoss , while the thin solid line is for the apr eos , but it is calculated including the contribution from the exchange of longitudinal plasmons alone .
one can see , that the inclusion of transverse plasmons lowers @xmath1 by a factor of three at @xmath422 g @xmath12 . with the fall of temperature
this lowering is stronger .
the exchange of transverse plasmons has not been included in previous calculations of the shear viscosity in neutron star cores , which has resulted in an overestimation of @xmath1 .
the viscosity @xmath1 for the pal ii eos ( the dotted line ) goes significantly higher than other curves due to a larger amount of protons ( and , therefore , electrons and muons ) for this eos . of neutrons times @xmath423 versus density for four eoss ( @xmath424 ) .
, scaledwidth=55.0% ] figure [ fig : viscn ] demonstrates the density dependence of the neutron shear viscosity multiplied by squared temperature , @xmath425 .
this combination is temperature independent .
the curves are calculated assuming the nucleon effective masses @xmath424 . in principle , the effective masses can be taken from microscopic calculations of an eos ; they can depend on @xmath9 , and our expressions for the shear viscosity allow one to incorporate this density dependence . here
we assume density independent effective masses by way of illustration . in fig .
[ fig : viscn ] , for simplicity , we do not present the results for the pal iv eos ; they are very close to the apr results .
one can see , that the neutron viscosity for the selected eoss differs within a factor of @xmath426 . at @xmath427 k. the curve cl is the approximation of cutler and lindblom @xcite of the results @xcite .
the curves bv ( 2bf ) and bv ( 2bf+3bf ) are obtained by benhar and valli @xcite for pure neutron matter taking into account two - body and two - body plus three - body forces , respectively .
other curves are our results for the viscosity @xmath428 , limited by neutron - neutron collisions alone , or for the viscosity @xmath4 , limited by neutron - neutron and neutron - proton collisions ; these curves are calculated for the apr eos assuming various phenomenological density independent effective masses @xmath252 and @xmath253
. , scaledwidth=70.0% ] figure [ fig : bv ] demonstrates the viscosity approximation of cutler and lindblom @xcite ( curve cl ) versus @xmath9 at @xmath427 k. recall that the approximation is based on the calculations by flowers and itoh @xcite performed for the eos of baym , bethe and pethick @xcite assuming in - vacuum nucleon - nucleon scattering cross sections and @xmath429 . also , we show self - consistent calculations of @xmath2 by benhar and valli @xcite for a pure neutron matter with the eos that is basically similar to apr ( with another version for three - nucleon interaction ) .
the authors have used one and the same nucleon interaction potential to derive the eos and @xmath2 .
the curve bv ( 2bf ) is their result ( from their fig .
1 ) obtained employing two - body nucleon forces ; the curve bv ( 2bf+3bf ) is obtained employing the two - body and three - body forces .
the effective mass @xmath430 is calculated self - consistently as a function of @xmath9 ( @xmath430 is different for both curves and , unfortunately , is not reported in @xcite ) .
all other curves in fig . [ fig : bv ] are our results for the apr eos assuming various values of @xmath430 and @xmath168 . for simplicity , these phenomenological values are taken density independent .
we show either the viscosity @xmath224 , limited by neutron - neutron collisions alone ( dashed lines ) , or the viscosity @xmath2 , limited by neutron - neutron and neutron - proton collisions ( solid lines ) .
one can see that the contribution of neutron - proton collisions is relatively small , while the dependence of the viscosity on nucleon effective masses is important .
smaller effective masses strongly increase the neutron viscosity . in the limit of @xmath429
we obtain the viscosity @xmath2 which is a factor of @xmath431 smaller than cl .
using the results of bhy01 for the thermal conductivity of neutrons @xmath432 , derived in the same approximations as our results for @xmath4 , we obtain the values of @xmath432 a factor of 24 smaller than those given by flowers and itoh @xcite .
the nature of this systematic disagreement of our results with the results of flowers and itoh is unclear .
we have checked that it can not be attributed to using different eoss .
a comparison of our results with those of benhar and valli @xcite is complicated because benhar and valli do not present the values of @xmath433 which they obtained for a neutron matter .
if , however , we take a reasonable value of @xmath434 , we obtain @xmath224 ( not shown in fig .
[ fig : bv ] ) very close to the curve bv ( 2bf ) of benhar and valli . in order to reproduce their bv ( 2bf+3bf ) curve with our equations
, we should employ a density dependent @xmath433 .
it should vary from @xmath435 at @xmath436 g @xmath12 to @xmath437 at @xmath438 g @xmath12 .
let us note , that the inclusion of three - nucleon interactions does reduce @xmath433 , and the reduction increases with density @xcite .
however , since we do not know exact values of @xmath433 , used in ref .
@xcite , we can not analyze the relative importance of @xmath433 and in - medium corrections to the squared matrix element .
we assume ( sec . [ viscn_norm ] ) that the effect of the effective masses is more important .
notice , in addition , that the in - vacuum differential neutron - neutron scattering cross section in ref .
@xcite ( the solid line in their fig .
3 ) seems underestimated .
k ( @xmath439 ) .
, scaledwidth=50.0% ] in fig .
[ fig : partial ] we compare partial shear viscosities in a neutron star core with the apr eos at @xmath418 k. previously , it has been widely thought that @xmath2 completely dominates over @xmath1 in the core of a nonsuperfluid star .
now we have considerably lowered both viscosities ( figs.[fig : viscem ] and [ fig : bv ] ) . the main contribution to the total shear viscosity
( @xmath440 , the solid line ) at @xmath418 k comes from the electrons ( @xmath54 , the dashed line ) .
the neutron viscosity @xmath2 ( the dash - dotted line ) is lower than @xmath54 .
note , however , that the relation between @xmath2 and @xmath54 is temperature - dependent ; when @xmath46 decreases , @xmath2 becomes more important ( see fig . [
fig : visct ] and a discussion below ) . the dotted line in fig .
[ fig : partial ] shows the muon shear viscosity @xmath55 . for @xmath418
k , it becomes comparable with @xmath2 at @xmath441 g @xmath12 .
now we discuss the shear viscosity in the presence of proton superfluidity ( superconductivity ) but for nonsuperfluid neutrons . for illustration , we take @xmath424 throughout a neutron star core .
k ) for the apr eos at @xmath442 g @xmath12 ( @xmath439 ) .
curves sf correspond to the superfluid case , while other curves are for normal matter .
the thin solid curve is @xmath1 calculated including the exchange of longitudinal plasmons alone.,scaledwidth=50.0% ] figure [ fig : visct ] demonstrates the temperature dependence of @xmath1 ( solid lines ) and @xmath2 ( dashed lines ) in the presence of proton superfluid ( @xmath443 k , curves sf ) and for non - superfluid matter ( unmarked curves ) at @xmath444 g @xmath12 . in a non - superfluid matter , @xmath1 exceeds @xmath2 at @xmath445 k ( for the adopted values of @xmath252 and @xmath253 ) but the situation reverses at lower @xmath46 . the reversal is a consequence of the different temperature behaviors , @xmath3 [ eq . ( [ eq : etaperp ] ) ] and @xmath446 .
the thin solid line shows @xmath1 calculated taking into account the exchange of longitudinal plasmons alone .
it demonstrates the standard fermi - system behavior , @xmath447 , and overestimates @xmath1 . at @xmath448
k the overestimation is small .
it reaches a factor of @xmath167 three at @xmath449 k , and exceeds one order of magnitude at @xmath450 k. proton superfluidity noticeably increases @xmath1 at @xmath451 and restores the fermi - liquid temperature behavior , @xmath447 [ eq .
( [ etaper_sf ] ) ] .
the increase of @xmath2 is not large because it comes from superfluid suppression of neutron - proton collisions which give a relatively small contribution to @xmath2 . in the presence of proton superfluidity
, @xmath1 completely dominates over @xmath2 . the electron shear viscosity in superfluid matter has recently been considered by andersson et al .
we have already discussed their approach in sec .
[ red_prot ] . for their eos and superfluidity model ,
their results overestimate @xmath54 , typically , by a factor of three in superfluid matter and by more than one order of magnitude in non - superfluid matter .
( left ) and the bulk viscosity @xmath452 ( right ) at different temperatures ( the values of @xmath453 are given near the curves ) in a neutron star core with the pal i eos ( see text ) .
solid curves are for normal matter ( n ) . dashed curves are for @xmath418 k and proton superfluidity with @xmath454 k. the viscosity @xmath452 is plotted for a neutron star vibrating at a frequency of @xmath455 s@xmath456 . , scaledwidth=100.0% ] figure [ fig : bulk ] compares the shear viscosity @xmath5 ( left panel ) with the bulk viscosity @xmath452 ( right panel ) determined by the direct and modified urca processes in the core of a vibrating neutron star at @xmath427 , @xmath457 , and @xmath458 k. the vibration frequency is set to be @xmath455 s@xmath456 ; these vibrations strongly affect @xmath452 but do not affect @xmath5 .
the bulk viscosity is calculated according to refs .
the eos is the same as in @xcite ( pal i with @xmath459 mev ) .
note , that @xmath5 decreases with growing @xmath46 , while @xmath452 increases ( e.g. , refs .
@xcite ) . for @xmath427
k , the shear viscosity dominates in the entire stellar core , while for @xmath418 k the bulk viscosity @xmath452 in the inner core ( where the direct urca process is allowed , after the jump of @xmath452 in the right panel ) becomes @xmath16710 times higher than @xmath5 . for @xmath448
k , the bulk viscosity completely dominates in the entire core .
the presence of proton superfluidity enhances @xmath5 and suppresses @xmath452 .
the dashed lines in the left and right panels of fig .
[ fig : bulk ] show @xmath5 and @xmath452 , respectively , in superfluid matter with @xmath454 k at @xmath418 k. superfluidity makes the shear viscosity more important .
note that shear perturbations in dense matter ( e.g. , associated with differential stellar rotation ) are damped by the shear viscosity and can be unaffected by the bulk viscosity . therefore
, the shear viscosity can be important for applications even if it is lower than the bulk viscosity .
( @xmath453 [ k ] = 7 , 8 , and 9 solid , dashed , and dash - dotted lines , respectively ) .
the left vertical dotted line shows the neutron drip density , the right line is the crust - core interface.,scaledwidth=50.0% ] finally , fig .
[ fig : corecrust ] compares the shear viscosity in the crust and the core of a neutron star . the viscosity in the crust
is calculated for cold - catalyzed matter @xcite using the results of refs .
@xcite and @xcite .
the former paper is devoted to the viscosity mediated by electron - ion collisions , while ref.@xcite deals with the contribution of electron - electron collisions taking into account the exchange of transverse plasmons . in the crust ,
the latter effect is not large .
we use the apr eos in the core , and the core is assumed to be non - superfluid .
the solid , dashed , and dash - dot lines correspond to @xmath427 , @xmath457 , and @xmath458 k , respectively .
the viscosity jump at the star crust - core interface ( at @xmath460 g @xmath12 ) is due to the disappearance of atomic nuclei in the core .
the nuclei , present in the crust , lower the viscosity owing to a very efficient electron - ion scattering .
we have calculated the shear viscosity in a neutron star core as a sum of the electron - muon viscosity @xmath1 and the neutron viscosity @xmath2 . calculating the viscosity @xmath1 , which is mediated by collisions of charged particles , we have taken into account the exchange of transverse plasmons ( that has not been done before ) .
our results include also the effects of proton superfluidity .
they are universal , presented in the form of analytic fit expressions convenient for implementing into computer codes for any eos of nucleon matter in neutron star cores .
our main conclusions are : 1 .
the exchange of transverse plasmons strongly reduces @xmath1 for all temperatures and densities of interest in a non - superfluid core . a low temperatures
, we have @xmath3 .
the viscosity @xmath1 generally dominates over @xmath2 , although @xmath2 can exceed @xmath1 at @xmath450 k and @xmath461 g @xmath12 ( for @xmath462 ) .
the viscosity @xmath4 strongly depends on the nucleon effective masses . typically , it is more than one order of magnitude lower , than that calculated in ref .
@xcite and parametrized in ref .
strong proton superfluidity significantly increases @xmath1 and restores its fermi - liquid temperature dependence , @xmath447 . in this regime
, @xmath1 exceeds @xmath2 .
the shear viscosity @xmath5 is comparable with the bulk viscosity @xmath452 at @xmath463 k ( for a star vibrating at a frequency @xmath464 s@xmath456 ) and dominates at lower @xmath46 .
superfluidity increases the importance of @xmath5 in comparison with @xmath452 .
our results can be used in simulations of neutron star hydrodynamics , in particular , to analyze the damping of internal differential rotation , stellar oscillations , gravitational wave driving instabilities .
our results can be improved further in many respects .
it would be most important to study the shear viscosity problem in the presence of neutron and proton superfluidity in the frame of multifluid hydrodynamics as discussed in sec .
[ superflu ] .
nevertheless , even our restricted standard one - fluid formulation is incomplete .
our calculations of @xmath4 can be improved by taking into account the medium effects on the matrix elements of nucleon - nucleon scattering .
we rely on the results of ref . @xcite that these medium effects are weaker than the effects of nucleon effective masses ( which we include explicitly ) .
an account for the medium effects on the matrix elements would complicate the expressions for @xmath4 ( making them non - universal ) . we have also neglected the effects of strong magnetic field which can modify the shear viscosity . for not too high magnetic fields , @xmath465 g , which do not affect the plasma polarization functions ( e.g. , ref .
@xcite ) , the generalization of the present results to the magnetic case is straightforward . for stronger fields ,
the polarization tensor becomes anisotropic and the viscosity problem is very complicated .
the present results are in line with our studies of kinetic properties of relativistic plasma taking into account the exchange of transverse plasmons .
these effects were studied by heiselberg and pethick @xcite for ultra - relativistic quark plasma .
they should be included in all calculations of kinetic properties of relativistic plasmas , particularly in neutron stars .
for the neutron star crust , the effect was studied in @xcite ( thermal conductivity ) and @xcite ( shear viscosity ) . for neutron star cores ,
it was analyzed in @xcite ( thermal conductivity ) , @xcite ( electrical conductivity ) , and @xcite ( neutrino pair emission in electron - electron collisions ) .
we are very grateful to u. lombardo , a. i. chugunov and d. a. baiko for useful discussions and critical remarks .
this work was partly supported by the dynasty foundation , by the russian foundation for basic research ( grants 08 - 02 - 00837 , 05 - 02 - 22003 ) , and by the state program `` leading scientific schools of russian federation '' ( grant nsh 2600.2008.2 ) .
here we present explicit expressions of the angular integrals @xmath466 , defined by eq .
( [ e : parint ] ) , for different values of @xmath467 , @xmath468
|
we calculate the shear viscosity @xmath0 in a neutron star core composed of nucleons , electrons and muons ( @xmath1 being the electron - muon viscosity , mediated by collisions of electrons and muons with charged particles , and @xmath2 the neutron viscosity , mediated by neutron - neutron and neutron - proton collisions ) . deriving @xmath1 , we take into account the landau damping in collisions of electrons and muons with charged particles via the exchange of transverse plasmons .
it lowers @xmath1 and leads to the non - standard temperature behavior @xmath3 .
the viscosity @xmath4 is calculated taking into account that in - medium effects modify nucleon effective masses in dense matter .
both viscosities , @xmath1 and @xmath2 , can be important , and both are calculated including the effects of proton superfluidity .
they are presented in the form valid for any equation of state of nucleon dense matter .
we analyze the density and temperature dependence of @xmath5 for different equations of state in neutron star cores , and compare @xmath5 with the bulk viscosity in the core and with the shear viscosity in the crust .
|
many quantum computing applications @xcite and quantum communication protocols @xcite , such as quantum teleportation @xcite , are based on the sharing of entangled two - photon states between two distant receiving stations .
the paired photons can be distributed through free space transmission channels , or alternatively , through single - mode fiber optical links . indeed
, fiber optical links are the most practical way to share entanglement between a large number of users .
the polarization degree of freedom is the most widely used resource to generate entanglement between two distant parties .
polarization - entangled photons can be generated by means of spontaneous parametric down - conversion ( spdc ) , a nonlinear optical process in which two lower - frequency photons ( signal and idler ) are generated when a strong pump interacts with the atoms of a nonlinear material .
subsequently to the generation , each photon constituting a pair can propagate through a different single - mode fiber , with the initial degree of entanglement between the photons being preserved over long distances @xcite .
one application that is attracting a lot of interest due its potential key role in the newly emerging quantum communication networks is the multi - user quantum key distribution ( qkd ) @xcite . in order to implement a multi - user qkd network , various frequency channels
can expediently be employed for transmitting individual entangled pairs . in this way , one can re - route on demand specific channels between users located in different sites of the optical network .
similar schemes , considering the emission of photon pairs in different spectral and spatial modes , have been presented in @xcite for an on - demand single - photon source based on a single crystal . to prepare polarization - entangled paired photons in many frequency channels at the same time , one needs to engineer an spdc process with an ultra - broad spectrum .
usually type - i or type-0 configurations are preferred . with the type - ii phase - matching ,
the two down - converted photons have different polarizations and consequently different group velocities , which reduces dramatically their bandwidth .
for instance , the fwhm bandwidth of an spdc process in a type - ii periodically - poled ( pp ) ktp crystal at @xmath0 nm is given by @xmath1 , where @xmath2 is the length of the crystal @xcite .
for @xmath3 mm , the bandwidth is @xmath4 nm . on the other hand , in a type-0 ppln configuration with the same crystal length @xmath3 mm , lim et al .
@xcite achieved an approximate tenfold increase of the bandwidth @xmath5 nm . even though one can always reduce the length of the nonlinear crystal in a type - ii configuration to achieve an increase of the bandwidth , this results in a reduction of the spectral brightness of the source .
alternatively to short bulk crystals , bragg reflection waveguides ( brws ) based on iii - v ternary semiconductor alloys ( al@xmath6ga@xmath7as ) offer the possibility to generate polarization - entangled photons with an ultra - large bandwidth . the most striking feature of the use of brw as a photon source is the capability of controlling the dispersive properties of all interacting waves in the spdc process , which in turn allows the tailoring of the bandwidth of the down - converted photons : from narrowband ( @xmath8 nm ) to ultra - broadband ( hundreds of nm ) @xcite , considering both type - i and type - ii configurations .
therefore , one can design a type - ii spdc process in brws with a bandwidth typical for type - i or type-0 processes .
the utilization of bragg reflection waveguide ( brw ) has further advantages over other conventional spdc sources due to the large nonlinear coefficient of semiconductors ( @xmath9 pm / v ) @xcite , broad transparency window ( @xmath10 - @xmath11 m ) and mature fabrication technologies that can be used for integration of the source of entangled photons with a light source and other optical elements .
additionally , a laser based on brws has already been demonstrated @xcite , opening a door for the integration in a single chip of the pump light source together with the brw , where the down - converted photons are generated @xcite .
al@xmath6ga@xmath7as is an optically isotropic semiconductor , precluding birefringent phase matching .
however , the modal phase - matching of the interacting waves ( pump , signal and idler ) can be achieved by letting each wave propagate in a different type of mode supported by the waveguide .
for instance , the phase - matching can be successfully achieved if the pump propagates in the waveguide in a bragg mode , whereas the signal and idler photons propagate in total - internal - reflection ( tir ) modes @xcite .
in order to investigate the potential of the proposed design for generating wavelength - multiplexed polarization - entangled photon pairs over many frequency channels , let us examine the biphoton generation in a collinear type - ii phase - matching scheme in the bragg reflection waveguide ( see fig .
[ fig1 ] ) .
a continuous - wave tm - polarized pump beam with frequency @xmath12 illuminates the waveguide and mediates the generation of a pair of photons with mutually orthogonal polarizations ( signal : te polarization ; idler : tm polarization ) .
the frequencies of the signal and idler photons are @xmath13 and @xmath14 , respectively , where @xmath15 is the degenerate central angular frequency of both photons , and @xmath16 is the angular frequency deviation from the central frequency .
the signal photon ( te ) propagates as a tir mode of the waveguide with spatial shape @xmath17 and propagation constant @xmath18 .
the idler photon ( tm ) , also a tir mode , has a spatial shape @xmath19 and propagation constant @xmath20 .
the pump beam is a bragg mode of the waveguide with spatial shape @xmath21 and propagation constant @xmath22 . at the output face of the nonlinear waveguide
, the quantum state of the biphoton can be written as @xcite @xmath23 is defined @xmath24 ^ 2 \gamma^2}{16 \pi \epsilon_0 c^3\,n_s(\omega_0 )
n_i(\omega_0 ) n_p(\omega_p ) } \right]^{1/2}.\ ] ] @xmath25 is the flux rate of pump photons , @xmath26 is the overlap integral of the spatial modes of all interacting waves in the transverse plane , and @xmath27 are their refractive indices . the joint spectral amplitude @xmath28 has the form @xmath29 \exp \left\ { i s_k ( \omega )
l/2 \right\}. \label{eg3}\ ] ] the ket @xmath30 ( @xmath31 ) designates a signal ( idler ) photon that propagates with polarization te ( tm ) in a mode of the waveguide with the spatial shape u@xmath32 ( u@xmath33 ) and frequency @xmath34 ( @xmath35 ) .
the phase - mismatch function reads @xmath36 , and @xmath37 .
the function @xmath38 is proportional to the probability of detection of a photon with polarization te and frequency @xmath39 in coincidence with a photon with tm polarization and frequency @xmath40 .
after the waveguide , a wavelength demultiplexer is used to separate all @xmath41 frequency channels into coupled fibers leading to the users of the network .
the bandwidth of each channel is @xmath42 and their central frequencies are @xmath43 , where @xmath44 is the inter - channel frequency spacing and the letter u(l ) indicates the upper ( lower ) path ( see fig .
[ fig1 ] ) . after the demultiplexer
, the quantum state of the down - converted photons can be written as @xmath45 where @xmath46 designates the frequency bandwidth from @xmath47 to @xmath48 coupled into every single fiber .
since we are interested in generating polarization - entangled paired photons coupled into single - mode fibers , the signal and idler photons in the upper and lower paths are projected into the fundamental mode ( @xmath49 ) of the fiber .
the coupling efficiency between the signal and idler modes , and the fundamental mode of the single - mode fiber are given by @xmath50 and @xmath51 .
they yield a value of @xmath52 in the whole bandwidth of interest , showing a minimal frequency dependence .
all the modes are normalized so that @xmath53 for @xmath54 . neglecting the vacuum contribution in the final quantum state , normalizing and tracing out the frequency degree of freedom ,
the two - photon state can be represented by the following density matrix , where we use the conventional ordering of rows and columns as @xmath55 where @xmath56 with @xmath57}=\alpha_n+\beta_n=1 $ ] .
( a ) .(a ) parameters of the structure : @xmath58 - core thickness ; @xmath59 - thicknesses of the alternating layers of the bragg reflector ; @xmath60 - aluminium concentration in the core ; @xmath61 - aluminium concentrations in the reflector s layers ; @xmath62 - the refractive index in the core ; @xmath63 - refractive indices in the reflector s layers ; @xmath64 -the inverse group velocity of the signal ( idler ) photon .
the structure is optimized for the collinear type - ii spdc .
( b ) profile of the refractive index along the _ y_-axis of the brw . [ cols="^,^",options="header " , ] ( b ) [ fig_tab_1 ] in the waveguide structure considered , the pump wavelength is @xmath65 nm . the frequency spacing between channels is @xmath66 ghz and the bandwidth of each channel is @xmath67 ghz , which corresponds approximately to @xmath68 nm at @xmath69 nm . the channel width and the spacing between channels were chosen according to the typical values used in commercial wdm systems .
channel @xmath70 corresponds to the wavelength @xmath71 nm in the upper path and to @xmath72 nm in the lower path . in order to reach a high number of frequency channels ,
the brw structure must be designed in such way , so as to permit the generation of signal - idler pairs with an ultra - broad spectrum in the type - ii configuration .
this is achieved when the group velocities of the te and tm modes are equal @xcite , i.e. , @xmath73 .
the modes of the structure and its propagation constants are obtained as a numerical solution of the maxwell equations inside the waveguide using the finite element method @xcite .
the waveguide design has been optimized by a genetic algorithm according to the requirements .
the final brw design is composed of two bragg reflectors , one placed above and one below the core .
each reflector contains 8 bi - layers .
the sellmeier equations for the refractive indices of the layers were taken from @xcite .
table 1 summarizes the main parameters of the structure . of the biphoton generated in the bragg reflection waveguide for type - ii phase - matching ( @xmath74 ) .
] inspection of eq .
( [ sigma ] ) shows that the effective nonlinearity @xmath75 of the waveguide spdc process depends on the effective area ( @xmath76 ) , which is related to the spatial overlap of the pump , signal and idler modes . for the structure considered ,
the effective area exhibits only minimal frequency dependence in the bandwidth of interest and it is equal to @xmath77 @xmath78 .
despite the fact that the large effective area will reduce the strength of the interaction , the high nonlinear coefficient still results in an efficiency that is a much higher than for other phase - matching platforms in waveguides or in bulk media . the total emission rate @xcite can be expressed using eq .
( [ state1 ] ) as @xmath79 . for our brw , the emission rate is @xmath80 @xmath81 . for comparison , for a typical ppln waveguide ( type-0 )
similar to the one used in @xcite , we obtain @xmath82 @xmath81 . the intensity of the joint spectral amplitude , given by eq .
( [ eg3 ] ) , is displayed in fig .
even though we are considering a type - ii configuration , the width ( fwhm ) of the spectrum is a staggering @xmath83 nm . the degree of entanglement in each spectral channel can be quantified by calculating the concurrence @xmath84 of the biphoton @xcite .
the concurrence is equal to 0 for a separable state and to 1 for a maximally entangled state . for the density matrix of eq .
( [ state3 ] ) we obtain @xcite
@xmath85 so the degree of entanglement depends on the symmetry of the spectral amplitude @xmath86 , i.e. , if @xmath87 the concurrence is maximum .
( solid line ) , @xmath88 ( dashed line ) and @xmath89(dotted - and - dashed line ) as a function of the frequency channel . ]
figure [ fig4 ] shows the values of @xmath88 , @xmath89 and @xmath84 for the first 200 channels .
@xmath90 is reached for the first 179 channels . the decrease ( increase ) of the parameters
@xmath89 ( @xmath88 ) reflects the fact that for frequency channels with a large detuning from the central frequency , one of the two polarization components of the polarization entangled state , @xmath91 or @xmath92 , shows a greater amplitude probability . in this case ,
one of the two options predominates .
therefore , if the goal is to generate a quantum state of the form @xmath93 in a specific frequency channel with @xmath94 , one can always modify the diagonal elements of the density matrix with a linear transformation optical system , keeping unaltered the degree of entanglement .
the number of frequency channels available depends on the concurrence required for the specific application .
a good example is a linear optical gate relying on the interference of photons on a beam splitter @xcite
. this would be especially important for the implementation of quantum teleportation , where the fidelity of the protocol depends strongly on the spectral indistinguishably between polarizations of the entangled state @xcite . in fig .
[ fig5 ] we plot the number of frequency channels available as a function of the minimum concurrence required .
for instance , if we select only frequency channels with @xmath95 , we have at our disposal @xmath96 channels , while for @xmath97 this number is reduced to @xmath98 channels . in the implementation of the system considered here in a _
real _ fiber - optics network , the number of frequency channels available can be limited by several factors .
for instance , it can be limited by the operational bandwidth of the demultiplexer ( see fig.1 ) .
this device should be designed to operate with the same broad spectral range of the photon pairs generated in the brw waveguide . when using a large number of channels , inspection of fig .
[ fig2 ] shows that channels far apart from the central frequency will exhibit a lower brightness . in this case ,
spectral shapers or appropriately designed filters , should be used to flatten the emission spectrum , similarly to the case of broadband gain - flattened erbium doped fiber amplifiers ( edfa ) .
notwithstanding , this might introduce some losses in the generation process , especially when considering a large number of channels , deteriorating the flux rate of the source .
interestingly , a similar problem appears in the context of optical coherence tomography ( oct ) , where large bandwidths are required to increase the imaging resolution . in oct ,
spectral shapers are used to obtain an optimum ( gaussian - like ) spectral shape @xcite .
finally , we should mention that the generation of polarization - entangled photons with the large bandwidths considered here require a precise control the group velocities of the interacting waves , which in turn requires a precise control of the waveguide parameters : refractive index and layer widths .
the effective number of available frequency channels in a specific application is inevitably linked to the degree of control of the fabrication process . since both down - converted photons are propagating as tir modes , they are more resistant to fabrication imperfections . for example , a change of about 10@xmath99 in the aluminium concentration in the core will reduce the spectral bandwidth to @xmath100 nm .
notwithstanding , it has to be stressed that the phase - matching condition for interacting waves is highly sensitive to any fabrication imperfection , therefore any small change of the structural parameters will lead to a shift of the central ( phase - matched ) wavelength .
in conclusion , we have presented a new type of highly efficient waveguide source for generating polarization - entangled photon pairs for its use in multi - frequency qkd networks . in spite of being a type - ii spdc source ,
the achieved bandwidth is even larger than the bandwidth usually obtained with type - i or type-0 sources .
the key enabling factor that allows us to achieve high efficiency of the nonlinear process together with a bandwidth increase is the fact that we can use a longer nonlinear material in a type - ii configuration , while at the same time keeping the broadband nature of the spdc process through the appropriate design of the bragg reflection waveguide structure . even though _ conventional _
sources based on the use of more common nonlinear materials , such as ktp or linbo@xmath101 , might also generate entangled pairs of photons with a large bandwidth , brws based on algaas compounds offer two main advantages : an enhanced capability to tailor the general properties of the downconverted photons , and the possibility of integration of different elements ( pump source , nonlinear waveguide and diverse optical elements ) in a chip platform based on an already mature technology , which could pave the way for entanglement - based technologies in _ out - of - the - lab scenarios_.
this work was supported by project fis2010 - 14831 and fet - open 255914 ( phorbitech ) .
j. s. thanks the project fi - dgr 2011 of the catalan government .
this work has also supported in part by projects cost oc 09026 , cz.1.05/2.1.00/03.0058 of the ministry of education , youth and sports of the czech republic and by projects prf-2011 - 009 and prf-2012 - 003 of palack university .
|
we put forward a new highly efficient source of paired photons entangled in polarization with an ultra - large bandwidth .
the photons are generated by means of a conveniently designed spontaneous parametric down - conversion process in a semiconductor type - ii bragg reflection waveguide .
the proposed scheme aims at being a key element of an integrated source of polarization - entangled photon pairs highly suitable for its use in a multi - user quantum - key - distribution system .
99 a. serafini , s. mancini , and s. bose , `` distributed quantum computation via optical fibers , '' phys .
. lett . * 96 * , 010503 ( 2006 ) .
j. i. cirac , a. k. ekert , s. f. huelga , and c. macchiavello , `` distributed quantum computation over noisy channels , '' phys .
rev .
a * 59 * , 42494254 ( 1999 ) .
g. ribordy , j. brendel , j. gautier , n. gisin , and h. zbinden , `` long - distance entanglement - based quantum key distribution , '' phys .
rev .
a * 63 * , 012309 ( 2000 ) .
a. k. ekert , `` quantum cryptography based on bell s theorem , '' phys .
rev .
lett . * 67 * , 661663 ( 1991 ) .
m. hillery , v. buek , and a. berthiaume , `` quantum secret sharing , '' phys .
rev .
a * 59 * , 18291834 ( 1999 ) . c. h. bennett , g. brassard , c. crepeau , r. jozsa , a. peres , and w. k. wootters , `` teleporting an unknown quantum state via dual classical and einstein - podolsky - rosen channels , '' phys .
. lett . * 70 * , 18951899 ( 1993 ) .
t. s. humble , and w. p. grice , `` spectral effects in quantum teleportation , '' phys .
rev .
a * 75 * , 022307 ( 2007 ) .
h. hubel , m. r. vanner , t. lederer , b. blauensteiner , t. lorunser , a. poppe , and a. zeilinger , `` high - fidelity transmission of polarization encoded qubits from an entangled source over 100 km of fiber , '' opt .
express * 15 * , 78537862 ( 2007 ) .
t. e. chapuran , p. toliver , n. a. peters , j. jackel , m. s. goodman , r. j. runser , s. r. mcnown , n. dallmann , r. j. hughes , k. p. mccabe , j. e. nordholt , c. g. peterson , k. t. tyagi , l. mercer , and h. dardy , `` optical networking for quantum key distribution and quantum communications , '' new j. phys . * 11 * , 105001 ( 2009 ) .
a. l. migdall , d. branning , and s. castelletto , `` tailoring single - photon and multiphoton probabilities of a single - photon on - demand source , '' phys .
rev .
a * 66 * , 053805 ( 2002 )
. j. h. shapiro , and f. n. wong , `` on - demand single - photon generation using a modular array of parametric downconverters with electro - optic polarization controls , '' opt .
lett . * 32 * , 26982700 ( 2007 ) .
a. fedrizzi , t. herbst , a. poppe , t. jennewein , and a. zeilinger , `` a wavelength - tunable fiber - coupled source of narrowband entangled photons , '' opt .
express * 15 * , 1537715386 ( 2007 ) .
h. c. lim , a. yoshizawa , h. tsuchida , and k. kikuchi , `` wavelength - multiplexed distribution of highly entangled photon - pairs over optical fiber , '' opt .
express * 26 * , 2209922104 ( 2008 ) .
p. abolghasem , j. han , b. j. bijlani , a. arjmand , and a. s. helmy , `` continuous - wave second harmonic generation in bragg reflection waveguides , '' opt . lett .
* 34 * , 94609467 ( 2009 ) .
k. thyagarajan , r. das , o. alibart , m. micheli , d. b. ostrowsky , and s. tanzilli , `` increased pump acceptance bandwidth in spontaneous parametric downconversion process using bragg reflection waveguides , '' opt .
express * 16 * , 35773582 ( 2008 ) .
d. kang and a. s. helmy , `` generation of polarization entangled photons using concurrent type - i and type-0 processes in algaas ridge waveguides , '' opt . lett . *
37 * , 14811483 ( 2012 ) .
i. shoji , t. kondo , a. kitamoto , m. shirane , r. ito , `` absolute scale of second - order nonlinear - optical coefficients , '' j. opt .
soc . am .
b * 14 * , 22682294 ( 1997 ) .
b. j. bijlani , and a. s. helmy , `` bragg reflection waveguide diode lasers , '' opt . lett . * 34 * , 37343736 ( 2009 ) . r. horn , p. abolghasem , b. j. bijlani , d. kang , a. s. helmy , and g. weihs , `` monolithic source of photon pairs , '' phys .
rev .
lett . * 108 * , 153605 ( 2012 ) .
a. s. helmy , b. bijlani , and p. abolghasem , `` phase matching in monolithic bragg reflection waveguides , '' opt . lett . * 32 * , 23992401 ( 2007 ) .
j. p. torres , k. banaszek , and i. a. walmsley , `` engineering nonlinear optic sources of photonic entanglement , '' prog .
optics * 56 * , 227331 ( 2011 ) .
s. v. zhukovsky , l. g. helt , d. kang , p. abolghasem , a. s. helmy , and j. e. sipe , `` generation of maximally - polarization - entangled photons on a chip , '' phys .
rev .
a * 85 * , 013838 ( 2012 ) .
j. p. torres , m. hendrych , and a. valencia , `` angular dispersion : an enabling tool in nonlinear and quantum optics , '' adv . opt
. photon .
* 2 * , 319369 ( 2010 ) .
j. jin , _ the finite element method in electromagnetics , 2nd edition _ , ( wiley - ieee press , 2002 ) .
s. gehrsitz , f .
k. reinhart , c. gourgon , n. herres , a. vonlanthan , and h. sigg , `` the refractive index of al(x)ga(1-x)as below the band gap : accurate determination and empirical modeling , '' j. appl . phys . * 87 * , 78257837 ( 2000 ) .
a. ling , a. lamas - linares , and c. kurtsiefer , `` absolute emission rates of spontaneous parametric down - conversion into single transverse gaussian modes , '' phys .
rev .
a * 77 * , 043834 ( 2008 ) .
s. hill , and w. k. wootters , `` entanglement of a pair of quantum bits , '' phys . rev .
lett . * 78 * , 50225025 ( 1997 ) .
k. wootters , `` entanglement of formation of an arbitrary state of two qubits , '' phys .
rev .
lett . * 80 * , 22452248 ( 1998 ) .
t. yu , and j. h. eberly , `` quantum open system theory : bipartite aspects , '' phys .
rev .
lett . * 97 * , 140403 ( 2006 ) .
y kim , and w. p. grice , `` reliability of the beam - splitter based bell - state measurement , '' phys .
rev .
a * 68 * , 062305 ( 2003 ) .
p. p. rohde , and t. c. ralph , `` frequency and temporal effects in linear optical quantum computing , '' phys .
rev .
a * 71 * , 032320 ( 2005 ) .
w. drexler , `` ultrahigh - resolution optical coherence tomography , '' j. biomedical opt .
* 9 * , 4774 ( 2004 ) .
|
the numerical calculations implicated in the study of nuclear reactions between heavy ions have an extreme complexity .
some times it is advisable to build simple models that allow to obtain concrete images of the phenomena based on classical conceptions and made useful predictions . in this perspective , d. h.
e. gross @xcite proposed the paradigmatic piston - gas model .
however , our interpretation differs , not sustaining a friction process .
we propose in this letter a classical , thermodynamical approach to determine the interaction potential to which two colliding heavy ions ( with mass number @xmath1 above 40 ) are submitted , and its dependency relatively to their nuclear temperature @xmath2 .
the high number of nucleons present inside each nucleus allows a macroscopic treatment .
also , the high value of the relative momentum @xmath3 with @xmath4 @xmath5 ensures that the reduced wavelength to be @xmath6 fm , which implies that the wave packet associated to each ion is much smaller that the ion radius ( @xmath7 fm ) .
the collision reactions are irreversible and can be treated as a dissipative phenomena due to friction " that are supposed to be occurring at the moment of impact .
however , we introduce here a different perspective , based on previous work done by r. de abreu et al .
@xcite , consisting in the following assumption : there is no friction , but an irreversible phenomena with cause at the difference of pressure @xmath8 felt " by the nucleon gas when ions collide with a target when compared with the pressure that nuclear matter exert in their boundaries when in thermodynamical equilibrium @xmath0 : @xmath9 with @xmath10 and @xmath11 denoting , resp .
, the speeds of the piston and of the gas particles , supposed to collide in chaotical manner with the piston and in ( relatively ) great number .
it is known that the variation of energy due to the piston is given by @xmath12 notice that we are treating here the repulsive region of the potential when the ions are compressing each other , and distant from the repulsive core , @xmath13 , otherwise we must consider the repulsive coulomb potential as well . as a first approximation , we assume @xmath0 independent from @xmath14 , and we can immediately integrate eq .
[ eq2 ] from @xmath15 , with @xmath14 denoting a minimal approaching distance and @xmath14 a given distance from the center of the nucleus : @xmath16 where @xmath1 is the area of contact between the two nucleus and @xmath17 is the pressure exerted by the nuclear matter , dependent on the gas temperature @xmath2 , chemical potential @xmath18 , and the nucleons mass @xmath19 .
we will assume that the gas of nucleons is a fermi gas without interaction between its particles .
the chemical potential is given by ( see , e.g. , ref .
@xcite ) @xmath20 it can be shown that @xmath21 and the entropy is ( @xmath22 ) @xmath23 the pressure is straightforwardly obtained : @xmath24 and it can be rewritten using eq .
[ eq6 ] : @xmath25 we assume that @xmath26 .
[ eq7a ] shows that the resulting compression due to the heavy ion collision is equivalent to an amount of work @xmath27 , and generation of entropy due to the new arrangement of particles . from the above ,
we obtain the interacting potential @xmath28 keeping with the analogy between the system piston - gas and the collision between heavy ions , we assume that the the energy of the incident ion is given by @xmath29 it results @xmath30 with @xmath31 we can go a step further to obtain @xmath32 where @xmath33 denotes the mass of the heavy ion and @xmath34 is the distance of minimal approximation , corresponding to the moment when the ions collide inelastically , dissipating energy and ejecting nucleons in the process . and ( b ) @xmath35
. the touching point distance of two ions is @xmath36 fm , @xmath37 mev , @xmath38 .
the square dots are data representing the calculations obtained in the hartree - fock - bogoliubov approximation with skm@xmath39 parameter set of the skyrme force @xcite.,title="fig:",width=3 ] + the mathematical resolution of eq .
[ eq11 ] is complex since we do nt have a previous idea of what is the form of the interaction potential @xmath40 ( in fact , it is what we want to know ) .
however , we may assume that the total energy of the heavy ion is much higher that the interaction potential in the range considered ( until the minimal distance ) , @xmath41 ( relativistic heavy - ions collisions ) , in which case we obtain a more simplified integral equation : @xmath42 here , @xmath43(r - a),\ ] ] and @xmath44 in the relativistic limit , the mathematical solution of eq . [ eq12 ] is given by @xmath45 @xmath46 is a constant of integration that appears after converting eq .
[ eq12 ] to a first - order differential equation . in fig .
[ fig1 ] it is shown the calculations using eq .
[ eq15 ] for the ion - ion potential for reaction @xmath47 .
the touching point distance of two ions is @xmath48 fm for the set of three reactions shown in table [ table1 ] despite the different number of nucleons in each nucleus , @xmath49 mev , @xmath38 mev@xmath50fm , and @xmath51 @xmath5 .
notice that @xmath52 fm is the critical distance in remarkable agreement with the expected value @xcite .
in the woods - saxon spherical potential describing single - particle levels , @xmath53 represents the surface thickness ( e.g. , ref .
@xcite ) .
the comparison is made with several calculations obtained in the hartree - fock - bogoliuobov approximation with skm@xmath39 parameter set of the skyrme force @xcite and our theoretical approximation of eq .
[ eq15 ] with parameters shown in table [ table1 ] .
the advantage of the method proposed in this letter stays in the possibility to obtain an analytical form for the interaction potential , more easy , fast and precise to use .
the simple physical model also points out that what occurs between the colliding heavy ions is not a friction process , but simply an irreversible process due to the difference of pressures felt by the nucleon gas before and during the collisional process . .
the square dots are data representing the calculations obtained in the hartree - fock - bogoliubov approximation with skm@xmath39 parameter set of the skyrme force @xcite.,title="fig:",width=3 ] + .parameters for the interaction potential in nuclear heavy ions collisions .
potentials best fitted for @xmath54 @xmath5 . [ cols="^,^,^,^ " , ]
it was obtained the interaction potential between two heavy ions in the framework of a simple , classical , one - dimensional piston - gas model , where there is no friction , but an irreversible phenomena with cause at the difference of pressure @xmath8 felt " by the nucleon gas when ions collide with a target when compared with the pressure that nuclear matter exert in their boundaries when in thermodynamical equilibrium @xmath0 .
our parametrized potentials are in good agreement with other theoretical potentials , particularly with the hartree - fock - bogoliuobov approximation with skm@xmath39 parameter set of the skyrme force , and with experimental data .
|
a simple one - dimensional gas - piston kinetic model gives the interaction potential between two colliding heavy ions . in the frame of the classical , thermodynamical approach ,
the colliding heavy ions are not submitted to friction , but produces an irreversible phenomena with cause at the difference of pressure @xmath0 felt " by the nucleon gas when ions collide with the target when compared with the pressure that nuclear matter exert on their boundaries when in thermodynamical equilibrium , and offers a straightforward way to calculate interacting potentials .
[ multiblock footnote omitted ]
|
thermal ratchets or brownian motors can be viewed as controllers that act on stochastic systems with the aim of inducing directed motion through the rectification of fluctuations @xcite . in most cases ,
the system to be controlled is modelled as a collection of brownian particles undergoing langevin dynamics , and the control action that is , the rectification mechanism is implemented by applying random or deterministic time - dependent perturbations to the particles . in this context , one can distinguish , as is common in control theory @xcite , two types of ratchets : ( i ) _ open - loop _ ratchets , which are ratchets that apply a rectifying potential independently of the state of the system to be controlled ; ( ii ) _ closed - loop _ or _ feedback
_ ratchets , whose rectification action on a system has an explicit dependence on that system s evolution in time .
examples of open - loop ratchets include the flashing ratchet @xcite and the rocking ratchet @xcite .
an example of closed - loop ratchet based on the flashing ratchet was proposed in @xcite ( see also @xcite ) .
this feedback flashing ratchet could be implemented experimentally by monitoring colloidal particles suspended in solution and by exposing the particles to a saw - tooth dielectric potential as in @xcite , but with the potential turned on and off depending on particles state .
the feedback ratchet of @xcite has also been proposed as a mechanism to explain the stepping motion of a two - headed motor protein @xcite . in a more general context
, recent experiments have shown that information about the location of a macrocycle in a rotaxane an organic molecule with a ring threaded onto an axle can be used to induce direct transport away from thermal equilibrium @xcite .
the operation of such a molecular ratchet is information - dependent , as it relies on knowledge of the position of the ring .
the use of information is also relevant in other chemical and biological ratchet - like systems @xcite .
the main motivation for studying closed - loop ratchets is that they have the potential to perform better as rectifiers of motion than open - loop ratchets , thereby opening the possibility of improving the technological applications of ratchets .
our goal in this paper is to establish a quantitative comparison between closed- and open - loop ratchets that explicitly focuses on what distinguishes them , namely the use of information .
this is done in three steps using the feedback ratchet of @xcite as a case example .
first , we show how the information used by this ratchet can be quantified ( sec .
[ secmodel ] ) .
then we study how the performance of that ratchet , measured by the magnitude of the flux of particles that it induces , varies as a function of the amount of information used in the ratchet effect ( sec .
[ secres ] ) .
the results obtained are discussed in sec .
[ secdisc ] and compared with those obtained with the open - loop version of the flashing ratchet , which operates without information . in sec .
[ secdisc ] , we also discuss the performance of the feedback ratchet as a function of the correlation established between the controlled system and the controller , and briefly discuss other performance measures for feedback ratchets , including the power output and the thermodynamic efficiency . the results that we obtain are in the end specific to the feedback ratchet of @xcite , but the method that we describe , which is inspired from maxwell s concept of thermodynamic demons @xcite , information theory @xcite and the work of one of us @xcite , is general and can be applied to other feedback ratchets and other control systems .
the model of feedback ratchet that we study is constructed as follows @xcite .
we consider a system of @xmath0 particles with positions @xmath1 , whose evolution is described by a set of overdamped langevin equations : @xmath2 here , @xmath3 is the friction coefficient related to the diffusion coefficient @xmath4 through einstein s relation @xmath5 , and @xmath6 are gaussian white noises of zero mean satisfying the fluctuation - dissipation relations @xmath7 .
all the particles are subjected to the same potential force @xmath8 , derived from the following asymmetric , saw - tooth potential : @xmath9 which is made periodic by the condition @xmath10 ; see fig . [
figpotentials](a ) . finally , @xmath11 is a control parameter that switches the potential on ( @xmath12 ) or off ( @xmath13 ) . in the open - loop flashing ratchet , the value of @xmath11
is typically changed periodically in time , whereas in the feedback ratchet of @xcite , @xmath11 is set to 1 if the net force @xmath14 applied to the particles is positive ; otherwise , @xmath15 .
thus @xmath16 , where @xmath17 is the heaviside function .
this feedback control strategy is the best possible strategy for maximizing the average velocity of one particle , but not the best strategy when it comes to more than one particle .
this can be understood by noting that the system s dynamics tends to get trapped as @xmath18 , with the consequence that the particle flux decreases to zero in this limit @xcite . by contrast
, the open - loop ratchet induces a flux of particles which is independent of the number of particles . for the remaining ,
it is useful to picture the ratchet as a maxwell s demon @xcite which rectifies the motion of the brownian particles by repeatedly estimating the sign of @xmath19 , and by subjecting the particles to the on or off potential depending on the value of the sign measured .
when selecting the potential at a given time @xmath20 , the demon uses only the sign of @xmath19 or , equivalently , @xmath11 at time @xmath20 .
it does not use past information about @xmath19 , nor does it use any detailed information about the positions of the particles . accordingly
, what should be quantified as the relevant information used by the feedback ratchet is the information content or _ variability _ of @xmath11 , given by its entropy @xmath21 where @xmath22 represents the probability that @xmath19 is negative .
the information @xmath23 is measured in bits , and represents the _ average _ information content of @xmath11 in that it corresponds to the average number of bits needed to store the random outcomes of @xmath11 @xcite .
since our goal is to study the performance of the demon as a function of @xmath23 , we need to supplement our measure of information with a way to vary that information .
this is accomplished by introducing noise in the estimation of @xmath19 .
it should be said that noise is always present in control systems in the measurement step , in the transmission of the measurement information to the controller , or even in the control - actuation step .
moreover , adding noise to a ratchet model can provide an effective way of describing an imperfect feedback controlled ratchet , such as one plagued by time delays [ 16 ] .
here we assume that there is a noise in the estimation of @xmath19 , which leads the demon to wrongly estimate the sign of @xmath19 with a probability @xmath24 $ ] , thereby leading it to apply the wrong potential with probability @xmath25 .
thus , when @xmath26 , the demon inadvertently switches off the potential with probability @xmath25 , resulting in an effective on potential @xmath27 .
conversely , when @xmath28 , the demon inadvertently switches on the potential with probability @xmath25 , resulting in an effective off potential @xmath29 .
the combination of these two situations leads , in effect , to having the following `` noisy '' control strategy : @xmath30 from the point of view of information theory , the noisy measurement of the sign of @xmath19 is equivalent to a noisy transmission channel known as the binary symmetric channel @xcite .
the average amount of information transmitted through this channel is measured in terms of a quantity known as the mutual information ( see @xcite for a general definition of this quantity ) . in our case
, the mutual information can be calculated exactly ( see sec .
8.1.4 of @xcite ) , and has for expression @xmath31 where @xmath32 is , as in eq . , the binary entropy function , @xmath25 is the noise level , and @xmath33 is the probability that the corrupted sign of @xmath19 is negative . in terms of the probability @xmath22
that the _ actual _ sign of @xmath19 is negative , we have @xmath34 .
note that @xmath22 depends in general on the number @xmath0 of particles , the characteristics of the ratchet potential @xmath35 , as well as @xmath25 , so that @xmath23 is a function of all these parameters .
the noise model that we consider is such that the ratchet operates with maximum information when @xmath36 , in which case @xmath37 , and with minimum information ( @xmath38 ) when @xmath39 . to study the exact performance of the ratchet in and in between these two regimes
, we derive in this section the expression of the flux of rectified particles as a function of @xmath25 , and rewrite this expression as a function of @xmath23 using eq .
( [ info1 ] ) . both the cases of one particle and a few particles are considered . for a single controlled particle ,
the net force is simply @xmath40 , with @xmath41 the position of the particle .
recalling the form of the saw - tooth potential defined in eq . , we have that @xmath28 for @xmath42 , and @xmath43 for @xmath44 . as a result ,
the effective control parameter @xmath45 can be rewritten as @xmath46 from this expression , we obtain an exact analytical expression for the average flux @xmath47 of the particle by solving the fokker - planck equation associated with the effective force @xmath48 that derives from the effective control potential depicted in fig .
[ figpotentials](b ) .
the result in the stationary regime is @xmath49 with @xmath50 -e^{\pm(2p-1)v_0}p(1-a)-a(1-p ) \nonumber\\ e & = & a^2(1-p)^2(1-pv_0-e^{-pv_0 } ) \nonumber\\ & & + ap(1-a)(1-p)(1-e^{-pv_0})[1-e^{(1-p)v_0 } ] \nonumber\\ & & + p^2(1-a)^2[1-e^{(1-p)v_0}+(1-p)v_0],\end{aligned}\ ] ] in units where @xmath51 , @xmath52 , and @xmath53 .
we have checked that this result is correct by performing langevin simulations of the feedback ratchet . as seen in fig .
[ figflux](a ) , the flux is maximum for @xmath36 and decreases monotonically to zero as @xmath25 goes to 1/2 . to transform the expression of @xmath54 shown above into a function of @xmath23 , we need to invert the expression of @xmath23 shown in ( [ info1 ] ) to obtain @xmath25 as a function of @xmath23
this requires the expression of @xmath22 , which is obtained by integrating over the space interval @xmath55 $ ] the stationary distribution of the effective fokker - planck equation , @xmath56-pv_0\bigg\ } , \label{b } \end{split}\ ] ] with @xmath54 given in ( [ vp ] ) , and units @xmath51 , @xmath52 , and @xmath53 .
this inversion gives the exact result for @xmath57 versus @xmath23 , which is plotted in fig .
[ figflux](b ) .
unfortunately , we can not provide a closed - form expression of @xmath58 because @xmath59 seems to have no closed - form expression .
however , it is possible to derive a useful approximation of the exact numerical result reported in fig .
[ figflux](b ) .
indeed , we can expand @xmath60 to second order in @xmath25 around the minimum located at @xmath39 to obtain @xmath61 assuming that @xmath22 does not depend on @xmath25 . then , for small potentials , i.e. , @xmath62 , we have @xmath63 and @xmath64 from eq . ( [ vp ] ) , so that @xmath65 this approximation is a priori valid only in the regime where @xmath66 and @xmath62 , but fig . [
figflux](b ) shows that it is accurate over the whole range of @xmath23 even when @xmath67 . an additional benefit of eq .
( [ v_of_i_one ] ) is that it is an upper bound on @xmath68 versus @xmath23 for any value of @xmath23 and any potential height @xmath69 .
this follows because the right - hand side of ( [ smallpot ] ) is an upper bound on the exact result shown in eq .
( [ vp ] ) @xcite ; see fig .
[ figflux ] .
approximations similar to those given in ( [ smallpot ] ) and ( [ v_of_i_one ] ) can also be derived for the case where more than one particle is controlled by the feedback ratchet . in the case of a few particles ,
the net force has a distribution @xmath70 which can be approximated by a gaussian distribution , @xmath71 having a zero mean and variance @xmath72 this gaussian approximation is derived under two basic assumptions [ 6 ] : ( i ) the positions of the particles are statistically independent ; ( ii ) the probability of finding a particle in a negative force interval ( for example @xmath55 $ ] ) is @xmath73 .
it can be shown that these two assumptions are verified for small potential even in the presence of noise ( i.e. , @xmath74 ) .
using the approximation shown in , we can obtain an approximate expression for the average center - of - mass velocity @xmath75 as a function of the transmission error @xmath25 using the relation @xmath76 with @xmath77 given by eq . .
in our case , we find @xmath78 we have compared this results with langevin simulations and found good fits for small potentials . for @xmath36 , the results of @xcite for @xmath75 is recovered .
in addition , we have verified numerically that eq .
( [ flux_small_approx ] ) is an upper bound of the exact flux for any given potential height . to turn the expression in ( [ flux_small_approx ] ) into an expression involving the information @xmath23
, we use the approximation shown in ( [ p_of_i_app ] ) again to obtain @xmath79 with @xmath22 the probability of having a negative force in the few particle case .
this probability can be approximated as @xmath80 using the same assumptions as those involved in the gaussian approximation for @xmath70 .
the resulting approximation for @xmath75 is similar to the approximation derived for the one - particle ratchet in that it is a good approximation for small values of the potential height , in addition to being an upper bound on the flux for any given potential height .
the latter property was checked numerically .
the accuracy of the approximation for the probability @xmath22 was also checked numerically .
\(1 ) the two approximations shown in ( [ v_of_i_one ] ) and ( [ v_of_i_few ] ) express the performance of the ratchet demon as a function of the information @xmath23 that the demon gathers through the noisy measurement of @xmath11 .
overall , we see from these results that the flux induced by the demon is maximum when it has maximum information , i.e. , @xmath81 , and is zero when it has zero information .
this applies both for the one - particle and few - particle ratchets . in both cases
, we further have that the flux decreases monotonically as @xmath23 decreases , and that the flux is approximately proportional to @xmath82 . the proportionality constant
entering in this relation depends on the system s characteristics , and shows , in the case of a few particles , an @xmath0 dependence that has the effect of reducing the flux as the number of particles is increased .
this extra reduction of the flux is related to the fact that the fluctuations of the force have a smaller amplitude as @xmath0 grows .
the decrease of flux directly associated with the decrease of information can be explained by noting again that the on and off potentials are partially ` mixed ' or randomized by the noise .
this is particularly evident when @xmath39 , i.e. , when @xmath38 . in this case
, the demon has a completely random estimate of the sign of @xmath19 which is uncorrelated with its true sign ; hence @xmath38 . with the random value of the sign
, the demon then applies a random potential to the particles , thereby injecting the noise of the estimation back into the motion of the particles .
such a feedback of estimation noise is often encountered in real control systems , and can be counteracted in various ways .
the most common is to rely on past measurements of the controlled system to better estimate its actual state ( filtering ) @xcite . for our demon
, this would mean acting with memory of past measurements of the sign of @xmath19 , and error - correcting those measurements to avoid inferring the wrong value of the sign of @xmath19 .
\(2 ) the expressions for the center - of - mass velocity as a function of the noise level @xmath25 or of the information @xmath23 can be rewritten in terms of the correlation @xmath83 where @xmath84 is the real sign of the net force that the particles would feel if the potential were on , while @xmath85 is the value that the controller receives .
this correlation can be computed by using @xmath86 where @xmath87 is the joint probability that the real net force @xmath88 is positive ( @xmath89 ) or negative ( @xmath90 ) and that the controller receives a positive ( @xmath91 ) or negative ( @xmath92 ) net force @xmath93 .
this joint probability is easily computed knowing that @xmath94 is different from @xmath95 with probability @xmath25 .
thus we have @xmath96 , @xmath97 , @xmath98 and @xmath99 , where @xmath22 is the probability of @xmath94 being negative , so that @xmath100 using this relation in eqs .
( [ vp ] ) , ( [ smallpot ] ) and ( [ flux_small_approx ] ) , we obtain the expressions for the center - of - mass velocity as a function of the correlation @xmath101 .
in particular , for small potentials , we obtain @xmath102 for one particle , and @xmath103 for few particles . as for eqs . and , the expressions shown above are upper bounds for the center - of - mass velocity for all potential heights , which show that the flux performance is reduced when the correlation between the controlled system and the controller decreases .
this loss of correlation is always present in physical systems , and can be due to noise in the measurement of @xmath19 or in the transmission of this measurement to the controller . in this sense , the expressions shown in and can be thought of as describing a noisy feedback ratchet for which all the noise sources are effectively described by @xmath101 .
such an effective description in terms of @xmath101 can be used , in addition , to model other imperfect feedbacks , such as time - delayed feedbacks @xcite . in the latter case , @xmath104 , where @xmath105 is a positive time delay , implying a loss of correlation between the actual force @xmath106 and the applied force @xmath93 .
time delays are expected to be present in the experimental implementation of the feedback ratchet mentioned in the introduction , in which colloidal particles suspended in a solution are monitored and exposed to a saw - tooth dielectric potential .
the previous discussion can directly be applied to this situation by computing @xmath101 from the time series of @xmath19 and the delayed signal @xmath107 , allowing the use of eqs . and to predict the effects of time delays on the flux .
the present noisy feedback model could also be useful as an effective description of the operation of an imperfect feedback loop in other ratchet - like systems @xcite .
\(3 ) experimental implementations of feedback ratchets are unavoidably imperfect and noisy due to the aforementioned time delays and other experimental imperfections .
these real - world limitations can be modelled , to a first level of approximation , by an effective noise level @xmath25 acting at the level of estimation . with this in mind , one can use our noisy feedback ratchet model as a valuable effective model for estimating the improvement in flux that can be obtained in experimental implementations , such as those proposed in @xcite .
the authors of @xcite , for example , propose a feedback ratchet based on a scanning line optical trap . from the relevant parameters of their experimental set - up they found a probability of @xmath108 of calculating the wrong sign of @xmath19 , and an information content of about @xmath109 bits . using the results presented here
, they obtain that no more than @xmath110 of the maximum gain achieved by the feedback strategy can be observed for that real system .
the experimental realization of this system is currently under way @xcite .
\(4 ) in general , the flux generated by the open - loop ratchet is much smaller than the flux generated by the feedback ratchet @xcite . for the saw - tooth potential considered here ,
the _ optimal _ open - loop protocol generating the largest flux is the periodic protocol with on - potential time @xmath111 and off - potential time @xmath112 . for @xmath113 and @xmath114 ,
the optimal values of these times are @xmath115 and @xmath116 , in units where @xmath52 and @xmath51 , yielding @xmath117 .
for the one - particle case , we have by comparison @xmath118 when @xmath23 is maximum . for other values of @xmath23 ,
the previous results for one and for few particles state that the center - of - mass flux @xmath75 is upper - bounded by @xmath119 , where @xmath120 is a constant depending on the system s characteristics ; this upper - bound is also greater than the open - loop value for most values of @xmath23 .
therefore , we can write @xmath121 the feedback protocol that we have considered , which performs an instantaneous maximization of the center - of - mass velocity @xcite , is the optimal protocol that maximizes the flux in the one particle case for a noiseless channel ( @xmath36 ) .
we expect this protocol to give a flux close to the maximum possible value in the few particles case and in the presence of noise with a memoryless protocol ( note that protocols with memory can perform error correction ) .
therefore , we expect eqs . and to be upper bounds of the maximum flux that can be obtained with a memoryless closed - loop control protocol that uses an amount of information @xmath122 about the system .
similarly , the inequality shown in eq . is expected to set an upper - bound on the maximum improvement that can result from changing an open - loop protocol to a memoryless closed - loop protocol .
this , at least , is the case for one particle , as the instant maximization protocol is optimal when applied to one particle .
for the few particle case , we expect the inequality to hold , although it could be violated by protocols other than the one considered here , as these could potentially be more efficient .
\(5 ) we have focused here on proving an upper bound for the particle flux because this quantity can readily be measured in experimental realizations of feedback ratchets @xcite . in a recent paper @xcite , written after the present one
, an analogous upper bound was derived for the power output of a feedback ratchet , based on the results and techniques presented here .
the difference between the particle flux and the power output is that the latter quantity requires that we impose a constant load force against the flux so as to compute the work done against the load ; see @xcite for more details .
\(6 ) another important performance measure for ratchets is the efficiency @xcite . for the computation of this quantity
, it is important to note that feedback ratchets have an extra energy input compared to open - loop ratchets , related to the fact that information has an energy cost @xcite .
this energy cost , known as landauer s erasure cost because it is incurred when information is erased , effectively prevents maxwell s demon - type engines , like feedback ratchets , to have efficiencies greater than one ( as required by the second law of thermodynamics ) @xcite .
the calculation of this energy cost requires the computation of the mutual information between the controlled system and the controller , conditioned on the past history of the controller s evolution @xcite .
the role of the conditioning is to take into account the correlations between the measurements , and to avoid redundancies in the computation of the entropy reduction .
the conditioning is also consistent with the fact that the controller s measurement record , seen as blocks of bits , must be compressed before it is erased in order to minimize the erasure cost ; see ref .
@xcite .
\(7 ) we have not addressed the many particle case , i.e. , the case where the fluctuations in the net force are smaller than the typical values of the net force .
the reason for this omission is that , in the many particle case , the maximum increase of performance that results from changing the optimal open - loop protocol to a closed - loop protocol is negligible .
in summary , we have quantified the information gathered by a feedback control ratchet , and have derived analytical upper bounds , expressed as a function of the information , which establish limits on the difference between the flux of particles created by a closed - loop , flashing ratchet and the flux created by its open - loop version .
these bounds provide a direct evaluation of the performance of the feedback ratchet as a function of the information that it uses , and make more precise the idea that feedback ratchets act like maxwell s demons that use information about the state of particles to rectify the particles motion .
in addition , the analytic results found for the flux are useful in that they provide an effective description of a feedback flashing ratchet affected either by noise in the measurement process or other imperfections in the feedback mechanism .
this effective description is useful for predicting the results of experimental realizations of feedback ratchets , as any experimental realization is subjected to noises , delays , and other imperfections in the feedback .
this work was supported by mcyt ( spain ) ( grants bfm2003 - 02547/fisi , fis2005 - 24376-e , and fis/2006 - 05895 ) , and by the esf program stochdyn .
m.f . acknowledges support from ucm ( spain ) through the grant beca complutense .
was supported by nserc ( canada ) , the royal society , and hefce ( england ) .
|
we study a feedback control version of the flashing brownian ratchet , in which the application of the flashing potential depends on the state of the particles to be controlled .
taking the view that the ratchet acts as a maxwell s demon , we study the relationship that exists between the performance of the demon as a rectifier of random motion and the amount of information gathered by the demon through measurements . in the context of a simple measurement model ,
we derive analytic expressions for the flux induced by the feedback ratchet when acting on one particle and a few particles , and compare these results with those obtained with its open - loop version , which operates without information .
our main finding is that the flux in the feedback case has an upper bound proportional to the square - root of the information .
our results provide a quantitative analysis of the value of information in feedback ratchets , as well as an effective description of imperfect or noisy feedback ratchets that are relevant for experimental applications .
pacs : 05.40.-a , 89.70.+c , 02.30.yy keywords : ratchet ; brownian motor ; feedback ; closed - loop ; information .
|
one of the most fascinating complex adaptive systems in nature is the brain . despite its relatively simple basic units
the neurons the cooperative bebaviour of the interconnected neurons and their functional implications are only poorly understood .
the problem in investigating this system is not only its complexity , because e.g. the human brain consists of about @xmath0 neurons @xcite , but also its characteristic cycle structure which is known as action - perception - cycle .
the difficulty with the action - perception - cycle , which was already known to von uexkll in 1928 @xcite , is that a closed formulation of the problem has to include a coupled description of the brain and the environment because the actions of an animal are transformed by the environment to perceptions which are transformed by the brain to actions and so on . from this
it is also clear that neither the perceptions nor the actions occurring in the system are randomly generated . in this paper
we address the question : how is the learning dynamics of a neural network affected by different mechanisms for the selection of an action ?
because learning in neural networks is modulated by a learning rule for the modification of the synaptic weights one can ask more precisely , if the learning rule itself is concerned by the action - selection mechanism .
we approach this problem by comparing two different biologically motivated learning rules for neural networks .
the first was proposed by bak and chialvo @xcite and combines experimental findings of frey and morris @xcite about _ synaptic tagging _ with a global reinforcement signal which can be interpreted as a dopamin signal e.g. as in the experiments of otmakhova and lisman @xcite .
the second was introduced by the author @xcite and extends the ingredients above by the results of fitzsimonds @xcite about heterosynaptic _ long - term depression _
( ltd ) which can be qualitatively explained by our stochastic learning rule .
both learning rules are local in the sense that the information , which is used for the synaptic modification , is only provided by the neurons which enclose the synapse and hence can be interpreted as extentions to the classical hebbian learning rule @xcite . as problem
to be learned we choose the problem of timing , e.g. catching a ball , in a recurrent network topology which is generated by an algorithm of watts and strogatz @xcite .
this network class was chosen because the topology is generated in dependence of one parameter , the so called rewiring parameter , and allows to convert a regularly connected network continously in a random one .
recently of special interest was the regime between these two extrema , called small world networks , which could be brought in contact with experimental results about the neuroanatomic structure @xcite .
this paper is organized as follows . in section [ model_timing ]
we define our model .
section [ results ] demonstrates the practical working mechanism exemplified in learning the problem of timing in a recurrent neural network .
we compare the learning behavior of our learning rule @xcite with the learning rule of chialvo and bak @xcite in dependence of two different action - selection - mechanisms .
the paper ends in section [ conclusions ] with conclusions and an prospect on future work .
if one wants to investigate the learning dynamic of a neural network one has to define every item of table [ gen_sys ] which characterizes the entire system .
.[gen_sys]characterization of the entire system [ cols="^ , < " , ] one recognizes by comparison with table [ list_ens1 ] that the overall results are confirmed .
learning rule 4.b ) obtains always significantly better results than 4.a ) .
moreover , a direct comparison between the learning rules for asm i. and ii . reveals that learning rule 4.a ) seems to be unaffected by the action - selection mechanism whereas learning rule 4.b ) is clearly influenced . to quantify the dependence of the learning behavior of the action - selection mechanism we calculate the mean first - passage time @xmath1 from the simulation results obtained so far .
figure [ meanfpt_rs ] compares the results for learning rule 4.a ) and 4.b ) in dependence of the rewiring parameter @xmath2 and the patterns to be learned .
one can clearly see that the mean first - passage time for learning rule 4.b ) ( upper ( lower ) two curves correspond to @xmath3 ( @xmath4 ) ) is significantly reduced for asm ii .
( full lines ) whereas the results for learning rule 4.a ) are not affected ( middle curves correspond to @xmath4 ) .
this can be explained by the different structure of both learning rules .
learning rule 4.a ) possesses no memory with respect to the outcomings of past results but only a tagging mechanism for the neurons which were involved in the last signal processing step .
hence it can not detect the differences of the two action - selection mechanisms because they differ only in the order of the presented patterns but not in the overall presentation statistics .
this follows from the fact that learning the last pattern takes about @xmath5 of the first - passage time .
learning rule 4.b ) is due to the neuron counters @xmath6 different in this point .
the neuron counters are a memory for the outcomings of the past results and thus can detect the slight difference in the two action - selection mechanisms .
we think that this result is worth to be discussed in detail because it reveals some deep characteristics of animals which is normally neglected in investigations of neural networks .
the consequences of the results obtained above are not only that the learning rule of a neural network effects on the neural activity by synaptic changes and hence on the behavior of an animal which is common sense , but also that the reverse holds .
that means the actions of an animal influence the learning rule of its neural network .
this is caused by the stimuli generated by the animal s actions which are represented in the examples above as patterns which lead to a modulation of the neural activity in the network and hence to a modulation of the learning rule due to memory effects by the neuron counters .
this seems to be plausible because we do not choose our actions randomly but we choose them to learn something as fast as possible to survive .
moreover , it is not only plausible but also efficient to us the action - selection mechanism as source of information which is shown in figure [ meanfpt_rs ] .
hence our investigations lead not only to a bottom - up communication but also to a top - down communication between different system levels . in this respect
our learning rule with neuron counters is different to all other hebb - like learning rules which has been proposed as extentions to the classical hebbian rule @xcite which lack the ability of a memory because they can not be affected by action - selection mechanisms which differ not in the presentation statistics but only in the presentation order .
in this article we investigated the properties of our recently proposed stochastic hebb - like learning rule for neural networks .
we demonstrated by extensive numerical simulations that the problem of timing can be learned in different topologies of a neural network generated by the algorithm of watts and strogatz @xcite .
a comparison with the learning rule of chialvo and bak @xcite gave not only always significantly better results but revealed that our stochastic hebb - like learning rule can discriminate between different action - selection mechanisms with the same presentation statistics but different presentation order .
this difference forms a source of information and can positively effect the learning behavior due to the bidirectional communication between different system levels .
this effect was only recognized because we did not want to model the brain of an animal but its action - perception - cycle schematically depicted in table [ gen_sys ] where the brain is only one part of the entire system . in summary our stochastic hebb - like learning rule
is not only universal applicable in feedforward multilayer networks @xcite but also in a class of recurrent networks generated by @xcite as demonstrated in this article . together with its biological interpretation as qualitative form of heterosynaptic plasticity @xcite and its sensitivity to the presentation order of the patterns to be learned
we belief that our learning rule unites some crucial ingredients on the way of our understanding of the action - perception - cycle and hence of the brain .
we belief that only such an integrated ansatz can explain the functional working method of the entire system because its parts are coupled in a nonlinear or stochastic way .
|
we demonstrate that our recently introduced stochastic hebb - like learning rule @xcite is capable of learning the problem of timing in general network topologies generated by an algorithm of watts and strogatz @xcite .
we compare our results with a learning rule proposed by bak and chialvo @xcite and obtain not only a significantly better convergence behavior but also a dependence of the presentation order of the patterns to be learned by introduction of an additional degree of freedom which allows the neural network to select the next pattern itself whereas the learning rule of bak and chialvo stays uneffected .
this dependence offers a bidirectional communication between a neuronal and a behavioural level and hence completes the action - perception - cycle which is a characteristics of any living being with a brain .
|
heavy quark effective field theory ( hqeft ) has arguably been the fastest growing sector of elementary particle theory in the four years since its introduction @xcite .
this can be ascribed to two features : on one hand the spin and flavor symmetries are phenomenologically successful @xcite and , on the other hand , corrections to the symmetry limit @xmath0 are well defined and therefore calculable . given this state of affairs
, it seems appropriate to investigate the theoretical formalism of hqeft in all orders of the combined @xmath1 , @xmath2 expansion .
it is only a slight exaggeration that quantum field theory ( qft ) has no physical content and is just the most powerful tool known for the calculation of @xmath3-matrix elements @xcite .
the fields and the off - shell values of their green s functions are not observable and therefore all qfts which lead to identical @xmath3-matrices are physically equivalent .
we are therefore free to choose the calculationally most convenient set of fields @xcite , similar to the choice of appropriate coordinates in classical physics .
this observation has borne the extremely powerful concept of effective field theory ( eft ) @xcite which underlies almost all phenomenologically successful applications of qft .
it is however far from obvious how the general theorems @xcite guaranteeing the independence of the @xmath3-matrix from the choice of interpolating fields translate to the intricate formalism of renormalized perturbation theory @xcite .
nevertheless , the so - called equivalence theorem , which formalizes this independence , has been proved rigorously @xcite , using the renormalized quantum action principles @xcite .
we shall extend this proof to hqeft below . in order to avoid any misconceptions
, we should stress that the ( colored ) heavy quark fields @xmath4 , @xmath5 of hqeft are _ not _ interpolating fields for @xmath3-matrix elements in non - perturbative qcd .
this rle is played by the meson and baryon fields .
however , it seems instructive to define the @xmath3-matrix for free quarks as an intermediate step in order to show how the equivalence theorem works in a perturbative calculation .
this note is organized as follows : to fix our notation and to introduce the technical subtleties of higher order calculations , we briefly describe hqeft in section [ sec : hqeft ] .
the quantum action principles are introduced in section [ sec : qap ] .
we sketch the proof of the equivalence theorems for green s functions and @xmath3-matrix elements in sections [ sec : et4gf ] and [ sec : et4sm ] , respectively . as an application
, we discuss the reparametrization invariance of hqeft in section [ sec : reparam ] and clarify the rle of the classical equations of motion in section [ sec : eom ] . finally , we present our conclusions in section [ sec : concl ] and describe why hqeft is more subtle than the quantum mechanical foldy - wouthuysen transformation .
within the framework of perturbation theory , the heavy mass expansion is constructed by performing the following steps on the green s functions , i.e. , feynman diagrams of qcd : heavy quark loops are integrated out and replaced by their expansion around vanishing momenta entering the loop .
the momenta of the remaining heavy quark propagators ( those connected to external heavy quark lines ) are replaced by the residual momenta @xmath6 , @xmath7 , @xmath8 , with @xmath9 the pole mass of the heavy quark .
the antiparticle pole is eliminated by expanding the denominators around @xmath10 : @xmath11 although this procedure is legitimate for tree level amplitudes and for the regularized loop diagrams , the different high energy behaviour of the propagators @xmath12 vs. @xmath13 in general destroys the equality of the full and effective theory amplitudes after renormalization .
one has to correct for this by introducing matching contributions into the effective theory .
this amounts to calculating the differences of the full and effective theory amplitudes for any amputated one - particle irreducible ( 1pi ) diagram of the full theory and absorb them into a set of local operator insertions in the effective theory .
the matching works here order by order in the loop expansion because the subtraction of the expansion about zero residual momentum @xmath14 improves the infrared ( ir ) behavior of the integrands , exactly in such a way as to render the corrections ir convergent @xcite .
note that it is essential for @xmath15 to be timelike for this to happen . a tree level , the above procedure may be carried out in the functional integral of qcd by integrating over the lower components of the heavy quark field with respect to @xmath15 and thus constructing an effective field theory ( hqeft ) .
the result @xcite @xmath16 containing the transverse derivative @xmath17 is expressed in terms of the upper component field @xmath18\ ] ] and the sources of upper and lower components @xmath19 the latter are not needed for deriving @xmath3-matrix elements , but one might find it convenient to retain them in order to easily obtain the effective theory analogues of operator insertions .
matching contributions enter in the same way as the tree - level quark - gluon vertex , so they can by included into ( [ l - tree ] ) by introducing a generalized covariant derivative @xcite @xmath20 with projections @xmath21 the effective lagrangian which generalizes ( [ l - tree ] ) is @xmath22 where @xmath23 summarizes operators consisting only of light fields which are introduced by integrating out heavy quark loops in the full theory .
this effective lagrangian is valid to arbitrary ( but finite ) order in the @xmath24 and loop expansions .
one can easily reformulate hqeft with an arbitrary residual mass @xmath25 of the order @xmath26
@xcite by the change @xmath27 and the induced changes in @xmath28 . for generality
, we include this residual mass term in the heavy quark propagator .
it is useful to establish some notation for a concise presentation of the quantum action principles ( qap ) . for convenience
, we shall denote all fields , heavy and light , generically by a scalar field @xmath29 .
the generating functional of the renormalized green s functions is given by @xmath30_{\cal l } = \sum_{n=0}^\infty \frac{i^n}{n ! } \int dx_1\ldots dx_n
j(x_1 ) \ldots j(x_n ) \left\langle 0 \left\vert { \rm t}\left [ \phi(x_1)\ldots\phi(x_n ) \right ] \right\vert 0 \right\rangle_{{\cal l}(\phi)}\ ] ] and we use the following shorthand for local operator insertions :
@xmath31_{\cal l } = \sum_{n=0}^\infty \frac{i^n}{n ! } \int dx_1\ldots j(x_1 ) \ldots \left\langle 0 \left\vert { \rm t}\left [ { \cal o}(z)\phi(x_1)\ldots \right ] \right\vert 0 \right\rangle_{{\cal l}(\phi)}.\ ] ] by including appropriate source terms into the lagrangian , we can express the generating functionals formally as functional integrals @xmath32_{\cal l } & = & \int{\cal d}\phi\ ; e^{i\int{\cal l}(\phi;j ) } \\ \label{eq : oz - pi } { \cal o}(z ) \downarrow z[j]_{\cal l } & = & \int{\cal d}\phi\ ; { \cal o}(z)\ , e^{i\int{\cal l}(\phi;j)},\end{aligned}\ ] ] but these expressions have to be renormalized properly . using formal manipulations of the functional integrals ( [ eq : z - pi ] ) and ( [ eq : oz - pi ] )
we can derive the three quantum action principles for the generating functionals immediately : 1 .
invariance under variations of the quantized fields @xmath29 , i.e. field equations : @xmath33_{\cal l } = 0 , \quad \delta { \cal l}(\phi ) = \frac{\delta { \cal l}(\phi)}{\delta \phi } \delta \phi\ ] ] and @xmath34 with @xmath35 an arbitrary local polynomial in @xmath29 and its derivatives .
2 . change under variation of external fields : @xmath36_{\cal l } = -i \frac{\delta z[j;\chi]_{\cal l}}{\delta \chi}(x)\ ] ] 3 .
change under variation of parameters in the lagrangian @xmath37_{\cal l(\eta ) } = -i \frac{\partial z[j]_{\cal l(\eta)}}{\partial \eta}.\ ] ] after renormalization the three qaps ( [ eq : qap1 ] ) , ( [ eq : qap3 ] ) , and ( [ eq : qap3 ] ) are valid without changes if the dimensional renormalization scheme is used @xcite . in other renormalization schemes
( e.g. bphz @xcite ) there are ( calculable ) normal product corrections @xcite .
since dimensional regularization is defined with the help of the schwinger parameter representation , which is also convenient for practical calculations , we define the parameter representation for diagrams containing heavy particles : any internal light line @xmath38 is assigned a parameter @xmath39 and an auxiliary vector @xmath40 .
the propagator is represented by @xmath41)\right|_{u_\ell=0}.\ ] ] where @xmath42 may contain algebraic objects like @xmath43 matrices , lorentz tensors etc .
any internal heavy line @xmath44 is assigned a parameter @xmath45 and an auxiliary vector @xmath46 .
the propagator is represented by @xmath47\cdot k_h - \beta_h[\delta m_h - i\epsilon])\right|_{u_h=0}.\ ] ] any vertex factor @xmath48 with @xmath49 , @xmath50 being the momenta of the attached external and internal lines , respectively , is translated into @xmath51 .
finally , the gaussian momentum integrations are formally carried out .
the requirement that no heavy particle loops are included ensures that all momentum integrations are indeed gaussian . in the following
we consider only the case of one heavy quark with velocity @xmath15 .
the generalization to the multiparticle sector is straightforward if the velocities of the heavy particles do not coincide .
here we shall _ not _ present a proof of theorem i of @xcite , i.e. the consistency of dimensional renormalization for hqeft in the parametric representation .
we do not expect a failure of a translation of the proof in @xcite , and we hope to come back to the purely technical details of such a proof in a later publication . in order to establish the first qap ( [ eq : qap1 ] ) in hqeft , we have to specify the separation of the lagrangian @xmath52 into the free ( @xmath53 ) and interaction ( @xmath54 ) parts .
we take @xmath53 to be equal to @xmath55 corresponding to the propagator ( [ eq : s+ ] ) .
the qap ( [ eq : qap1 ] ) then says @xmath56_{\cal l } = -\left(\bar h_v p(\phi)\,\frac{\delta{\cal l}_{\rm int}}{\delta\bar h_v}\right)(z ) \downarrow z[j]_{\cal l}\ ] ] where now @xmath29 denotes the light fields of the theory only . in order to remain in the one heavy particle sector ,
we insist that the field variations be linear in the heavy quark field . in diagrammatic language
, the content of ( [ eq : qap1-eff ] ) is that attaching the factor @xmath57 to a vertex is equivalent to contracting the neighbouring heavy quark line to a point . in the parametric representation
, we have to show that @xmath58 which is true since apart from a global factor @xmath59 , the dimensionally regularized integrand @xmath60 depends on @xmath46 and @xmath45 only in the combination @xmath61 .
the desired equality follows from the fact that @xmath62 since @xmath60 falls off exponentially at the upper limit .
the proof that the qap remains valid after renormalization proceeds exactly as in @xcite .
the only potential obstruction are diagrams such as depicted in fig . 1a , where the subdiagram enclosed by dashed lines has no equivalent if the center line of the diagram is contracted to a point ( cf .
also fig . 1 of @xcite ) .
the corresponding counterterm diagram would contribute to the left - hand side of ( [ eq : qap1-eff ] ) , but not to the right - hand side .
however , this counterterm diagram ( fig .
1b ) necessarily contains a heavy quark loop and thus vanishes identically .
the proof of the second qap ( [ eq : qap2 ] ) is simple combinatorics since @xmath53 is independent of the external fields .
the same applies to the third qap ( [ eq : qap3 ] , [ eq : qap3-eff ] ) if @xmath63 is independent of the parameter @xmath64 .
therefore we shall restrict ourselves to the case where the varied parameter is the velocity of the heavy quark @xmath65 the case @xmath66 can be treated similarly .
we have to prove : @xmath67_{\cal l } = - \int dz ( \bar h_v \delta v\cdot\partial h_v)(z ) \downarrow
z[j]_{\cal l } + i \frac{\partial{\cal l}_{\rm int}}{\partial \eta } \downarrow z[j]_{\cal l}.\ ] ] the second term on the right - hand side is cancelled exactly by the terms on the left - hand - side where the derivative acts on the vertices . furthermore , acting with @xmath68 on the propagator part of the amplitude results in terms of the form @xmath69,\beta\,\delta m,\ldots)\right|_{u=0 } } \nonumber\\ & = & \left.\int_0^\infty d\beta\,\beta\ , \delta v\cdot\frac{\partial}{\partial u } i(u+\beta[v+\eta\,\delta v],\beta\,\delta m,\ldots)\right|_{u=0 } \\
& = & \left.\int_0^\infty d\beta\,d\beta'\delta v\cdot \frac{\partial}{\partial u ' } i(u+\beta[v+\eta\,\delta v]+u'+\beta'[v+\eta\,\delta v],(\beta+\beta')\delta m,\ldots ) \right|_{\stackrel{\scriptstyle u=0}{u'=0}}.\nonumber\end{aligned}\ ] ] the last expression corresponds to a feynman diagram where two heavy lines are joined by the quadratic vertex @xmath70 , which is equivalently obtained from the first term of the right - hand side of ( [ eq : qap3-eff ] ) .
simple combinatorics completes the proof for the regularized amplitudes , and the validity of ( [ eq : qap3-eff ] ) after renormalization is proved as described in @xcite .
the equivalence theorem asserts the following relations among green s functions @xmath71 \right\vert 0 \right\rangle_{{\cal l}(\phi ) } = \left\langle 0 \left\vert { \rm t}\left [ \phi'(x_1)\ldots\phi'(x_n ) \right ] \right\vert 0 \right\rangle_{{\cal l'}(\phi)={\cal l}(\phi')}\ ] ] of fields which are related by a reparametrization @xmath72 in all orders of a simultaneous expansion in the numbers of loops and powers of @xmath64 .
we concentrate on field reparametrizations which are linear in the heavy field @xmath5 , but may be nonlinear in the other fields of the theory : @xmath73 the structure of the heavy quark lagrangian allows to partially integrate so that the derivatives in @xmath74 act on @xmath4 and the light fields only .
this simplifies the notation in the following arguments .
of course , the proof can be applied to @xmath4 also .
simultaneous changes in @xmath4 and @xmath5 may be splitted into subsequent reparametrizations of @xmath4 and @xmath75 .
we want to show @xmath76 \right\vert 0 \right\rangle_{{\cal l}(\bar h_v , h_v ) } = \left\langle 0 \left\vert { \rm t}\left [ \bar h_v'(x)\ldots \right ] \right\vert 0 \right\rangle_{{\cal l}'(\bar h_v , h_v ) = { \cal l}(\bar h_v',h_v)}.\ ] ] where it is understood that the green s functions are evaluated to some finite order @xmath77 .
following lam @xcite , we define @xmath78_{{\cal l}(\bar h_v + \eta \bar h_vf(\phi,\partial),h_v)}.\ ] ] the first qap ( [ eq : qap1 ] ) tells us that @xmath79 for any local polynomial @xmath35 in the light fields . from ( [ eq : hv - prime ] ) this is equivalent to @xmath80 after iterating this equation @xmath81 times by replacing @xmath35 by @xmath82 for @xmath83 on the left - hand side , we arrive at @xmath84 since we are interested in the variation of @xmath85 with @xmath64 , we use the third qap ( [ eq : qap3 ] ) to get @xmath86 applying ( [ eq : eta - n ] ) with @xmath87 , we see that the generating functional @xmath85 is independent of @xmath64 in any finite order @xmath81 : @xmath88 taking derivatives with respect to the sources completes the proof of the equivalence theorem ( [ eq : eqt - gf ] ) for renormalized green s functions in the effective theory .
operator insertions as those responsible for weak decays may also be considered by coupling them to some source .
we emphasize that this proof holds to finite order in @xmath64 and in the dimensional renormalization scheme ( ms or @xmath89 ) only .
in other renormalization schemes ( e.g. bphz , lattice ) there are corrections to ( [ eq : eqt - gf ] ) .
the same is true for resummed variations which modify the propagator .
after discussing the proper definition of @xmath3-matrix elements in hqeft , we shall derive the equivalence theorem for these @xmath3-matrix elements . in perturbation theory ( or for that matter in any non - confining theory like qed ) , there are asymptotic states corresponding to single heavy particles . translating the familiar lsz reduction formula for a massive fermion in momentum space @xmath90 \right\vert 0 \right\rangle}\right|_{p^2=m^2}\ ] ] to the effective theory , and setting @xmath91 , we use @xmath92 to arrive at @xmath93 \right\vert 0 \right\rangle } \right|_{v\cdot k = -k^2/2m}. \nonumber\end{aligned}\ ] ] here the kinematical factor @xmath94 follows from the matching condition ( [ ubar ] ) for the heavy particle wave functions .
alternatively , we can derive it from the residue of the heavy particle two - point function @xmath95 \right\vert 0 \right\rangle}$ ] at the one heavy particle pole .
the additional term @xmath96 in ( [ lsz - eft ] ) cancels the multiple poles in the heavy - quark green s functions that arise from insertions of the @xmath24 terms in the lagrangian into external lines .
having defined the @xmath3-matrix for heavy quark states , the proof that the @xmath3-matrix is unchanged by reparametrizations , i.e. , that the @xmath3-matrix elements calculated from the original hqeft are the same as those calculated with the new lagrangian obtained by the change of variables ( [ eq : hv - prime ] ) , proceeds exactly as in ordinary quantum field theory @xcite .
the additional terms in the green s functions on the right - hand side of ( [ eq : eqt - gf ] ) contribute in such a way that the residue of the pole at @xmath97 ( which appears as a series of multiple poles in the effective theory ) is multiplied by some function of @xmath98 identical for all green s functions of the heavy quark .
this function only modifies the factor @xmath99 in ( [ lsz - eft ] ) and therefore does not contribute to @xmath3-matrix elements . any contribution which has no pole at @xmath100 vanishes after applying ( [ lsz - eft ] ) . in the real world
heavy quarks come in bound states , and the @xmath3-matrix is defined in terms of those . denoting a heavy ( scalar ) bound state with mass @xmath101 generically by @xmath102 ,
a typical matrix element is given by @xmath103 \right\vert x \right\rangle \right|_{p^2=m^2 } \nonumber\end{aligned}\ ] ] where @xmath104 \right\vert 0 \right\rangle}\right|_{p^2=m^2}\ ] ] and @xmath105 is any operator such that @xmath106 does not vanish .
in the effective theory we have @xmath107 \right\vert x \right\rangle \right|_{v\cdot k + \frac{k^2}{2 m } = \frac{m^2-m^2}{2 m } } , \nonumber\end{aligned}\ ] ] where @xmath108 is some interpolating field of the effective theory , and the definition of @xmath109 is modified analogously @xmath110 \right\vert 0 \right\rangle \right|_{v\cdot k + \frac{k^2}{2 m } = \frac{m^2-m^2}{2m}}.\end{aligned}\ ] ] the difference of hadron and quark masses has to be treated in the same perturbative expansion @xmath111 so that multiple poles in the matrix element due to the mass shift are cancelled separately in any order of the @xmath24 expansion . of course
, one could rearrange the series to be an expansion in @xmath112 or any other physically sensible mass by suitable choice of @xmath9 .
using similar arguments as for the free quark case , it becomes obvious that any reparametrization of the heavy quark field that changes the form of the interpolating field @xmath105 has no effect on the ratio ( [ lsz - b - eff ] ) .
since the choice of @xmath105 has been arbitrary from the beginning , one might as well keep the definition of @xmath105 fixed while calculating a matrix element using the reparametrized lagrangian .
the equivalence theorem ensures that masses and other properties of bound states remain unchanged .
however , since nonperturbative qcd can not be formulated using the ms or @xmath89 scheme , in order to calculate a particular matrix element one first has to match renormalization schemes ( e.g. lattice versus @xmath89 ) . in a general scheme
the equivalence theorem does not hold without corrections , so one will in general obtain different matching contributions in different versions of the effective theory .
as an application of the qaps and the equivalence theorem , we discuss the implication of the luke - manohar reparametrization invariance @xcite on the effective lagrangian . for simplicity ,
we keep @xmath25 equal to zero in this section .
an infinitesimal change in the velocity @xmath15 used in the derivation of ( [ l - loop ] ) @xmath113 causes a corresponding change in the field @xmath4 @xmath114 since the @xmath115 field is integrated out , this implies @xmath116 inserting this variation into the effective lagrangian ( [ l - loop ] ) , a straightforward calculation shows that @xmath117 is reparametrization invariant @xmath118 if and only if @xmath119.\ ] ] assuming gauge invariance , this condition is satisfied if and only if the generalized covariant derivative @xmath120 depends on the ordinary covariant derivative @xmath121 and the velocity @xmath15 in the combination @xmath122 this is exactly what one would expect since @xmath123 is the full momentum of the heavy quark , the quantity that enters into the matching calculation . in particular , since @xmath124 reparametrization invariance is valid on tree level , as it has been shown already in @xcite .
the qaps imply that the relations following from this type of reparametrization invariance remain valid after renormalization , if the division of the effective lagrangian into renormalized part and counterterms is carried out using the ms or @xmath89 renormalization scheme . in other schemes
reparametrization invariance may be broken .
the physical content of the first qap ( [ eq : qap1 ] , [ eq : qap1-eff ] ) is that naive application of the classical equations of motion @xmath125 is justified in the @xmath89 renormalization scheme , where @xmath126 denotes the terms of order @xmath24 and higher in the effective lagrangian .
however , as it stands the qap applies only to single operator insertions such as weak currents .
if multiple operator insertions @xmath127 are considered , the easiest way to obtain the correct terms is to couple sources @xmath128 to them and include them in the effective lagrangian . in higher orders of the loop expansion ,
1pi diagrams containing multiple insertions generate counterterms nonlinear in the sources , so that the effective lagrangian is some polynomial in @xmath128 . employing the equivalence theorem , one can look for a reparametrization of the heavy quark field that results in a particular simplification .
if the lagrangian contains an operator of the form @xmath129 one may set @xmath130 so that the variation of the lowest - order term @xmath131 eliminates @xmath132 in favor of a tower of new higher - dimensional operators .
this method may be used to eliminate all terms containing @xmath133 from the interaction lagrangian . however
, the familiar luke - manohar reparametrization invariance is obscured in this process .
furthermore , the kinematical normalization factor @xmath99 in ( [ eq : z(k ) ] ) has to be recalculated , but this is only relevant for free heavy quarks as external states .
in fact , any allowed reparametrization is of the form ( [ eq : rep - gen ] ) and thus equivalent to the application of the equations of motion : the additional operator introduced into the lagrangian is @xmath134 the equivalence theorem tells us that the vanishing of this operator works for an arbitrary number of insertions , whereas the qap applies only to single insertions . on the other hand ,
reparametrizations are as well possible in the full theory , if the additional terms introduced in the qcd lagrangian are treated perturbatively . since the effective theory may be constructed also from the reparametrized lagrangian by calculating the appropriate matching corrections , we have the following situation : the equivalence theorem in the full _ and _ effective theories , respectively , guarantees that the matching contributions do not depend on in which theory reparametrizations are carried out , if we use dimensional renormalization in both cases . in ordinary field theory one usually considers only field redefinitions that are relativistically invariant @xmath135 such redefinitions may be mirrored in the effective theory ; we observe that they preserve the luke - manohar reparametrization invariance of the effective lagrangian .
other transformations correspond to lorentz non - invariant reparametrizations of the full theory .
in this letter we have established the quantum action principles ( qaps ) in hqeft which justify formal manipulations of the functional integral . using the qaps , we have proved the equivalence theorem for the effective theory and thus have shown under what conditions field redefinitions in hqeft are viable .
a particular class consists of transformations corresponding to the lorentz invariance of qcd the luke - manohar reparametrization invariance of hqeft . in practice
, the equivalence theorem serves as a guideline how the classical equations of motion manifest themselves once radiative corrections and renormalization are taken into account . from our discussion
it should have become clear that unless one is willing to handle normal product corrections it is necessary to use a ms - like renormalization scheme , and to make sure that the free lagrangian ( i.e. , the propagator ) is not affected . in particular
, this applies to attempts to derive the heavy quark lagrangian employing the foldy - wouthuysen transformation familiar from single - particle quantum mechanics @xcite . by its very nature
this transformation modifies the quark propagator in the full theory and thus violates the hypothesis of the equivalence theorem .
since the authors of @xcite do not discuss how to calculate the necessary normal product corrections , the argument given in @xcite is not a complete derivation of hqeft . on the other hand ,
the lagrangian provided in @xcite can be reduced to the tree - level lagrangian ( [ l - tree ] ) by a series of field redefinitions which are compatible with the equivalence theorem of the _ effective _ theory - matrix elements in the full theory ( [ lsz - ft ] , [ lsz - b ] ) as well as in he effective theory ( [ lsz - eft ] , [ lsz - b - eff ] ) . ] .
thus beyond tree - level the foldy - wouthuysen approach can _ only _ be justified if it is amended by some prescription to calculate matching corrections , which will then yield the same results as the conventional approach @xcite .
to summarize , the form of the hqeft lagrangian is by no means unique , but as long as a ms - like renormalization scheme is used , different versions can be mapped onto each other by perturbative field redefinitions . however , in any renormalization scheme where the qaps hold in the weak sense only ( i.e. , normal product corrections arise ) , one has to perform a new matching calculation if one wants to carry out reparametrizations . in particular , in a general scheme the classical equations of motion can not be naively applied , and reparametrization invariance is lost beyond leading order in the @xmath136 and @xmath24 expansions . 10
* potential obstruction in the proof of the qap .
the open square denotes the operator insertion on the left - hand side of ( 23 ) .
|
we discuss the quantum action principles and equations of motion for heavy quark effective field theory .
we prove the so - called equivalence theorem for hqeft which states that the physical predictions of hqeft are independent from the choice of interpolating fields .
en passant we point out that hqeft is in fact more subtle than the quantum mechanical foldy - wouthuysen transformation .
|
nonlinear topologically nontrivial excitations ( solitons ) are well known to play a special role in low dimensional magnetic systems .
for example , the presence of vortices in 2d easy plane ( ep ) magnets gives rise to the berezinski kosterlitz
thouless ( bkt ) phase transition.@xcite kinks in 1d magnets and localized belavin polyakov ( bp ) solitons @xcite in 2d isotropic magnets are responsible for the destruction of long range order at finite temperature .
the soliton signatures in dynamical response functions can be observed experimentally .
translational motion of solitons leads to the so called soliton central peak.@xcite another possibility to detect soliton signature is to look for magnon modes , localized at the soliton ( _ local modes _ , lm ) ; such modes are the cause of soliton magnetic resonance at the characteristic frequencies of `` intrinsic '' motion.@xcite this can be explained within the scope of the so called soliton phenomenology , where the magnet can be described as a two component gas of elementary excitations : solitons and magnons . such an approach was developed for 1d magnets.@xcite it was shown that the contribution of magnons and soliton magnon interaction is important for this approach , as is obvious , for example , in the discussion of soliton magnetic resonance.@xcite but for 1d magnets the soliton
magnon scattering causes a change of the magnon density of states , which is necessary for a self consistent calculation of the temperature dependence of the soliton density.@xcite for the 2d case , the concept of a soliton ( vortex ) gas has been extended to describe ep magnets above the bkt vortex
antivortex unbinding transition , with a finite density of vortices.@xcite the concept of localized topological solitons ( @xmath4topological solitons , see details in ref . )
was used to explain the epr line width in easy
axis ( ea ) magnets.@xcite the classical example of a localized soliton is bp solution , which exists in isotropic magnets .
the scattering problem for such solitons allows an exact analytical solution because of the scale and conformal invariance of the model.@xcite for ea magnets it is necessary to take account of some important modifications due to the breaking of such symmetries . that is why , both for ea and ep 2d magnets , the soliton density has not been calculated , but has been used as an input parameter .
the general features of the 2d soliton dynamics are not clear at present .
in particular , the form of inertial terms in the dynamical equations for the soliton centre is unknown ; so the study of localized states is an issue of current research .
note that the problem of the soliton dynamics and the problem of the existence of lm are intimately connected to each other , and to the problem of soliton
magnon scattering .
for example , using numerical data for the scattering amplitude , a non newtonian effective equation of motion of the magnetic vortex was constructed.@xcite usually non1d solitons and , especially , the problem of soliton
magnon scattering , are treated numerically.@xcite only in the case of purely isotropic magnets , with the exact bp solution , the analytical investigation of the scattering problem has been done recently.@xcite in particular , it was found that a soliton has a number of local modes of zero frequency;@xcite such modes determine the main features of the scattering picture.@xcite however , even a small uniaxial anisotropy leads to principal changes .
truly local modes ( in contrast to quasi local ones , below ) with nonzero frequencies were found numerically in ref . for the soliton in ea ferromagnets ( fm ) for some special values of the azimuthal quantum number @xmath2 ;
namely the mode with @xmath5 for the soliton with the topological charge @xmath6 , and the mode with @xmath7 for @xmath8 were investigated .
note that local modes do not exist for the models of easy plane and isotropic ferromagnets , where the magnon spectrum has a gapless dispersion law .
the case of the 2d ep antiferromagnet is an exception ; for this model a finite frequency , truly localized internal mode with @xmath9 exists inside the continuum spectrum.@xcite recently there has also been renewed attention to the internal dynamics of the soliton .
it is caused by the studying of small ferromagnetic particles in the so called vortex state ( magnetic dots ) .
magnetic dots are micron sized magnetic samples , placed on a non
magnetic substrate ; they have different shapes : circular , quadratic , etc .
the magnetic dots are interesting for the usage in the high density magnetic storage devices@xcite and from a fundamental point of view . as the dots are rather large ( the size of magnetic dot exceeds the critical size of a single domain ) , their ground state could be inhomogeneous .
such a dot in the vortex state contains a magnetic singularity in the centre of the vortex , or the belavin
polyakov soliton type.@xcite the discrete magnon modes were observed for the uniformly magnetized dots by a resonance technique;@xcite their theory was constructed using the landau lifshitz equations@xcite . in this paper
we consider the magnon modes which exist in 2d heisenberg ea fm with a topological soliton .
the model of an easy axis ferromagnet is presented in sec .
[ sec : model ] .
an approximate solution for the soliton structure is examined . in sec .
[ sec : modes ] we formulate equations , describing magnons and solitons . in sec .
[ sec : local ] truly local magnon modes are calculated analytically by variational approach for the limiting cases of large and small radius solitons , and numerically for arbitrary radii . in sec .
[ sec : scattering - numerical ] the problem of soliton
magnon scattering is formulated and treated numerically for different partial modes .
the phenomenology of a pseudopotential is proposed in sec .
[ sec : pseudopotential ] to analyze analytically both local states and the scattering problem . in sec .
[ sec : eq - mot ] we use results for local modes to derive the equations of soliton motion .
such equations are of newtonian type ( with the account of gyroscopic force ) , in contrast to both the case of the bp soliton in the isotropic magnet@xcite and the case of the vortex in the ep magnet@xcite . in the framework of this equation the soliton
effective mass is calculated . a discussion and concluding remarks are presented in sec .
[ sec : conclusion ] .
the dynamics of the classical ferromagnet is described by the landau
lifshitz equations for the normalized magnetization @xmath10 . in angular variables , @xmath11
, these equations correspond to the lagrangian @xmath12 with the energy density @xmath13 , @xmath14 where @xmath15 , @xmath16 is the exchange integral , @xmath17 the atomic spin , @xmath18 the characteristic scale ( `` magnetic length '' ) , @xmath19 the energy of the anisotropy , and @xmath20 the spin wave stiffness .
the model of ea fm has well - known magnon excitations @xmath21 , @xmath22 above the homogeneous ground state .
the dispersion law for the magnons has a finite activation frequency @xmath23 , @xmath24 where @xmath25 is the wave vector .
the simplest nonlinear excitation in the system is the dynamical ( precessional ) topological soliton@xcite @xmath26 where @xmath27 and @xmath28 are polar coordinates in the plane of the magnet , the integer @xmath29 plays the role of a topological charge , and @xmath30 ( @xmath31 ) is the frequency of the internal precession .
the necessity to consider solitons with internal precession is caused by the fact that , according to the derrick hobart theorem , static solitons in models like ( [ eq : w ] ) are unstable .
formally , there is no function @xmath32 , which could provide a minimum of the energy @xmath33 the precessional soliton corresponds to the conditional minimum of the energy ( [ eq : energy ] ) for a given number @xmath34 of magnons bound in the soliton @xmath35 which is conserved for the uniaxial case ( @xmath36 is the characteristic number of bound magnons in the 2d soliton),@xcite see also refs . .
this corresponds to the condition @xmath37=0 $ ] .
the form of the function @xmath32 is defined by the ordinary differential equation , @xmath38 with the boundary conditions ( [ eq : dyn_soliton(1 ) ] ) .
the precession frequency @xmath30 is fully determined by the number of bound magnons @xmath34 .
the dependence @xmath39 was calculated numerically in ref . ; large radius solitons ( @xmath40 ) have a very low precession frequency ( @xmath41 ) ; and @xmath42 for small radius solitons ( @xmath43 ) . in the case of large radius solitons ,
@xmath44 , the approximate `` domain wall '' solution@xcite @xmath45 is applicable , which describes a 1d domain wall at the distance @xmath0 from the origin .
one can obtain the relation between the soliton radius @xmath0 and the frequency of the soliton precession @xmath30 from variational condition , in which @xmath0 plays the role of a fitting parameter for the function ( [ eq : theta_r>>1 ] ) . in the first approximation @xmath46 .
the opposite case of small radius solitons , @xmath3 , corresponds to the belavin
polyakov solution @xmath47 , which describes the isotropic limit of the model ( [ eq : l ] ) ; it is valid for distances @xmath48.@xcite at larger distances the algebraic decay is replaced by the general exponential form , @xmath49 the dependence @xmath50 was calculated both analytically and numerically in refs . .
the frequency of the soliton precession satisfies @xmath51 when @xmath52 , but the dependence @xmath39 is not analytic , and @xmath53 as @xmath54.@xcite to describe approximately the soliton structure in the intermediate case of arbitrary @xmath0 , we use a fitting method.@xcite we choose a trial function of the form @xmath55,\ ] ] which has valid asymptotics for @xmath56 and @xmath57 ( we discuss the case @xmath6 only ) .
the parameter @xmath58 can be found numerically by fitting the trial function ( [ eq : theta - trial ] ) to the numerical solution of eq .
( [ eq : dyn_soliton(2 ) ] ) .
such an approximation shows that the trial function describes the soliton shape with an accuracy of about @xmath59 , see table [ table : fitting ] . from this table
one can see that the values of the fitting parameter @xmath58 are very close to the `` exact '' soliton radii ; it should be understood by the fact that @xmath60 .
numerical data for @xmath32 and fitting results for @xmath61 are plotted in fig .
[ fig : structure ] .
to analyze magnons on the soliton background , it is convenient to introduce local coordinates @xmath62 , which describe the configuration of the magnetization unit vector @xmath63 in the unperturbed soliton : @xmath64 coincides with @xmath63 in the soliton ( [ eq : dyn_soliton(1 ) ] ) , @xmath65 , and @xmath66 .
then the linear oscillations around the soliton can be described in terms of projections of the magnetization @xmath63 on the local axes : @xmath67 and @xmath68 . in the absence of the soliton , @xmath69 , @xmath70
, such oscillations correspond to free magnons in the rotating coordinate frame .
the linearized equations for @xmath71 and @xmath72 can be presented in the form : [ eq : vartheta&mu ] @xmath73 \vartheta + \frac{2q \cos\theta_0}{r^2}\frac{\partial \mu}{\partial \chi } = \frac1{\omega_0\delta_0 ^ 2 } \frac{\partial\mu}{\partial t } , \\ \label{eq :
mu } % & & \left [ - \nabla^2 + { \cal u}_2(r ) \right ] \mu - \frac{2q \cos\theta_0}{r^2}\frac{\partial \vartheta}{\partial \chi } = - \frac1{\omega_0\delta_0 ^ 2}\frac{\partial\vartheta}{\partial t}\end{aligned}\ ] ] with the `` potentials '' [ eq : u1&u2 ] @xmath74 where @xmath75 is the radial part of the laplace operator .
it is convenient to represent the solutions of eqs .
( [ eq : vartheta&mu ] ) in terms of a partial wave expansion for @xmath71 and @xmath72 [ eq : vartheta&mu - via - u&v ] @xmath76 where @xmath77 is a full set of eigennumbers , @xmath78 and @xmath2 being radial and azimuthal quantum numbers , respectively .
the quantity @xmath79 is the magnon frequency in the rotating frame ; @xmath78 and @xmath80 are connected by the dispersion law , cf .
( [ eq : dispersion_law ] ) , @xmath81 where @xmath82 is the magnon gap frequency in the rotating frame . below we
will replace @xmath80 by @xmath83 , almost everywhere without confusion , because we will work in the rotating frame mainly ; the index @xmath84 will be omitted , too .
thus we then finally obtain the following eigenvalue problem ( evp ) for the functions @xmath85 and @xmath86 : [ eq : u&v ] @xmath87 u = w(r ) v,\\ \label{eq : u&v(2 ) } % \hat{h}_2 u & \equiv & \left [ \hat{h}_0 + \frac{m^2}{r^2 } - v(r ) + \frac{{\omega}}{\omega_0\delta_0 ^ 2 } \right ] v = w(r ) u,\end{aligned}\ ] ] where @xmath88 is the 2d radial schrdinger operator ; the `` potentials '' are [ eq : potentials ] @xmath89 note , that the eqs .
( [ eq : u&v ] ) are invariant under the conjugations @xmath90 , @xmath91 , and @xmath92 . in a classical theory
we can choose any sign of the frequency ; but in order to make contact with quantum theory , with a positive frequency and energy @xmath93 ) for @xmath94 and @xmath95 .
besides there appears a difference between eqs .
( [ eq : u&v(1 ) ] ) and ( [ eq : u&v(2 ) ] ) . since ( [ eq : u&v(2 ) ] ) has the asymptotically equivalent form , @xmath96 , the function @xmath86 has an exponential behaviour ; at the same time eq .
( [ eq : u&v(1 ) ] ) yields oscillating solutions .
so , by choosing @xmath97 , the variable @xmath86 becomes a slave variable in the evp ( [ eq : u&v ] ) .
the `` coupling potential '' @xmath98 , @xmath99,\end{aligned}\ ] ] is positive , it has a maximum value at a distance of about the soliton radius @xmath0 from the origin , vanishing both at @xmath100 and far from the soliton . using the `` domain wall '' approximation ( [ eq : theta_r>>1 ] ) ,
one obtains that the maximum value of @xmath101 , when @xmath44 . in the opposite case of small @xmath0 ,
in the range @xmath102 the soliton looks like the belavin
polyakov solution ; here @xmath103 , and @xmath104 in the main exchange approximation.@xcite therefore , one can suppose that @xmath105 for any @xmath0 .
in fact , this is confirmed by numerical calculations , see fig .
[ fig : w ] .
however , one can see below in sec .
[ sec : local ] that the problem is impoverished when the `` coupling potential '' @xmath106 is ignored ; even in the limits of small ( large ) soliton radii , the main properties of local modes vanish , if we assume that @xmath107 .
the reason is that a set of more than one coupled schrdinger like equations has properties , which are absent from a single equation .
to explain this fact , remember that for a single 1d schrdinger equation there exist two independent solutions ; for the eigenfrequencies inside the continuous spectrum such solutions have oscillating asymptotics . for a set of 2 , 3 ,
equations there exist 4 , 6 ,
linearly independent solutions , respectively ; some of them can have exponential asymptotics .
naturally , the real modes have an oscillational form here ; we will use this fact below for the numerical analysis .
but it is necessary to take into account the exponential solutions , too . in particular , as was shown in ref .
, such exponential asymptotics could correspond to truly localized states inside the continuous spectrum , which are forbidden for equations of the schrdinger form . in the reduced evp with two independent equations like ( [ eq : u&v ] ) with @xmath104
, there is a continuous spectrum @xmath108 for the function @xmath109 ( @xmath110 , @xmath111 ) and @xmath112 for the function @xmath113 ( @xmath114 , @xmath115 ) .
there are no limitations for the lowest value of the discrete levels ; in particular , the frequencies of the discrete spectrum could be less then @xmath116 for the function @xmath85 . at the same time , as follows from the general set of coupled eqs .
( [ eq : u&v ] ) , frequencies of local modes lie in the region @xmath117 .
therefore , it is necessary to solve the general evp ( [ eq : u&v ] ) for an exact analysis of the problem .
we should note here that the model of an isotropic magnet is a special case , in which the `` coupling '' potential is exactly equal to zero . to study an evp like ( [ eq : u&v ] ) with @xmath104 , it is sufficient to take into account one equation only ( namely , the equation for the variable @xmath85 ) ; the second one , which has unphysical solutions for @xmath97 , was ignored in ref ..
that is why the results for ea fm , obtained in this article , and the results for the isotropic magnet@xcite differ strongly even for small anisotropy .
one difference is caused by the presence of truly local modes .
another one is due to the peculiar structure of the quasi continuous spectrum with @xmath118 for a finite geometry : the spectrum of the ea fm has a system of doublets with opposite signs of @xmath2 , while there are no doublets for the isotropic magnet .
according to the dispersion law ( [ eq : dispersion_law ] ) , free magnons only exist when @xmath108 .
it is naturally to look for local modes in the range @xmath119 ( remember that we discuss the case @xmath120 ) . in principle
, such local modes could exist inside the continuous spectrum , as for the case of the nonlinear @xmath121model , describing the ep antiferromagnet.@xcite however , our study shows that this is not the case for the ea fm .
existence of local modes in such a magnet was found numerically in ref . for a soliton with some special parameters ; namely the mode with @xmath5 for the soliton with the topological charge @xmath6 , and the mode with @xmath7 for @xmath8 were calculated .
such modes appear in field theories , which include soliton like solutions.@xcite to study local modes , note that near the soliton centre , @xmath48 , one can obtain the asymptotics of the magnon amplitudes in the following form : [ eq : u&v4x<<1 ] @xmath122 , \\ \label{eq : v4x<<1 } v_m(r ) & = & \varepsilon { \cal a}_m \left(\frac{r}{\delta_0}\right)^{|q+m| } \left[1+o\left({r^2}/{\delta_0 ^ 2}\right)\right].\end{aligned}\ ] ] here the coefficient @xmath123 can be determined by the normalization ; the presence of a nonzero factor @xmath124 is caused by the `` interaction '' @xmath106 between eqs .
( [ eq : u&v(1 ) ] ) and ( [ eq : u&v(2 ) ] ) ; its value can not be found through this asymptotic expansion . in the opposite case of large distance ,
@xmath125 , the evp ( [ eq : u&v ] ) has the following general solution ( here and below we consider the most interesting case of solitons with unit topological charge @xmath6 , which has lowest energy ) [ eq : u&v - loc - infty ] @xmath126 where @xmath127 and @xmath128 are macdonald and modified bessel functions respectively , and @xmath129 .
note , that at large distances , @xmath130 , the modified bessel functions have exponential behaviour , @xmath131 , @xmath132 , so we use the asymptotical condition @xmath133 to obtain the localized solutions .
firstly note that for the special cases @xmath134 there exist `` zero modes '' with @xmath135 .
such modes are caused by the internal symmetry of the problem .
they have the same form as for ep magnets.@xcite one of the zero modes , the so called translational mode with @xmath136 , [ eq : m=-1,0,+1 ] @xmath137 describes the position shift of the soliton .
another one with @xmath9 @xmath138 is a rotational mode , which corresponds to the presence of an arbitrary parameter , the phase @xmath139 , in the soliton structure ( [ eq : dyn_soliton(1 ) ] ) . for the isotropic magnet , as well as for the ep magnet ,
similar modes exist , too ; but there they are quasi local due to their slow power law decay , @xmath140 , or @xmath141 . in the ea magnets the decay has an exponential form .
the remaining modes could be investigated numerically ; analytical results were obtained for some limiting cases only , see below .
to study the evp numerically , we use the two
parameter shooting method for the integration of the set of eqs .
( [ eq : u&v ] ) ; a similar one
parameter shooting scheme was proposed in ref . to study the vortex
magnon scattering problem in 2d antiferromagnets .
numerically , we `` kill '' two growing exponents in ( [ eq : u&v - loc - infty ] ) by choosing two shooting parameters @xmath124 and @xmath83 . as a result ,
we have found numerically that there are a number of local modes . modes with opposite signs of @xmath2 bind to doublets ; this picture is well
pronounced for @xmath142 in the limit of large @xmath0 , see fig .
[ fig : omega ] .
such local modes exist only for not too small soliton radii ( not for large @xmath30 )
. eigenfrequencies for such modes with arbitrary azimuthal numbers @xmath2 ( @xmath142 ) increase rapidly with the decreasing of the soliton radius , and reach the boundary of the continuous spectrum .
it looks as if all such modes `` leave ''
the region of the gap in the spectrum , if the soliton radius is smaller than some critical value @xmath143 , see table [ table : rc ] .
it is natural to suppose that all modes with @xmath142 transform to the quasi local ones with decreasing of @xmath0 ; this fact can be tested numerically due to the singularities of the scattering amplitudes on such quasi local modes , see the next section .
thus , in the range @xmath144 there is only one local mode with non zero frequency , @xmath5 , see fig .
[ fig : omega ] .
that is why only the mode with @xmath5 was observed in ref . for the soliton with @xmath6 .
the eigenfunctions of bound states @xmath85 and @xmath86 are localized near the soliton core .
the functions @xmath85 and @xmath86 are plotted in fig .
[ fig : local_r ] for the mode with @xmath5 ; the picture of well localized eigenfunctions is specially pronounced for the large radius solitons , see fig .
[ fig : local_r](a ) . in order to analyze the spectrum of local modes analytically ,
let us reformulate the evp ( [ eq : u&v ] ) as a variational problem for the functional @xmath145 = \int_0^\infty rdr\left(u \hat{h}_1 u + v \hat{h}_2 v - 2wuv \right).\ ] ] we start with large radius solitons , @xmath44 , when the `` domain wall '' approximation ( [ eq : theta_r>>1 ] ) is valid . in the region of interest , @xmath146 ,
the operator @xmath147 from the evp ( [ eq : u&v ] ) has in the main approximation for @xmath148 a simple reflectionless potential , @xmath149\ ] ] with a single localized state @xmath150 , corresponding to the zeroth eigenvalue @xmath151 the other potentials in the evp can be considered as corrections to @xmath152 ; this is the reason to put the trial functions in the form @xmath153 , @xmath154 with trial parameters @xmath155 and @xmath156 .
the variational approach with such simple trial functions leads to eigenfrequencies for the different modes in the form @xmath157 where @xmath158 . however , there is a difference between the real solutions @xmath85 , @xmath86 , and the trial function @xmath150 .
this difference provides , e.g. , the existence of `` zero modes '' .
we do not like to improve the form of the eigenfunctions , but only the coefficients in the eqs .
( [ eq : omega_m ] ) , which describe the eigenfrequencies . using the additional condition for `` zero modes '' ,
@xmath159 the eigenfrequencies @xmath160 can be expressed through two independent coefficients only .
it is convenient to present the eigenfrequencies in the form @xmath161 using trial functions @xmath150 , we can calculate the parameters @xmath84 and @xmath162 for the large radius solitons , [ eq : omega_m_cases ] @xmath163 in this case the spectrum has the form @xmath164 this formula corresponds very well to the numerical data for @xmath44 , see fig . [
fig : omega ] . in the main approximation for @xmath148
there is a system of doublets ( for all modes with @xmath142 ) with quadratic dependence of the mean frequency on @xmath148 , @xmath165 and the splitting is a small cubic correction , @xmath166 .
for the special case @xmath5 the dependence @xmath167 has the form @xmath168 eq . ( [ eq : omega_m(r>>1 ) ] ) describes qualitatively all modes with arbitrary @xmath0 . in particular
, it describes the fact , mentioned above , that modes reach the boundary of the continuous spectrum at some finite frequency of the soliton precession @xmath169 .
let us estimate the positions of the crossover points , where the modes `` leave '' the discrete spectrum .
using eq .
( [ eq : omega_m(r>>1 ) ] ) , one can obtain the condition for the critical values of @xmath170 : @xmath171 .
we expect that the local modes transform into the quasi
local ones inside the continuous spectrum ; but their frequencies can still be described by the formula ( [ eq : omega_m(r>>1 ) ] ) , see sec .
[ sec : pseudopotential ] . for the small radius solitons , @xmath3 ,
there only exist local modes with @xmath172 .
thus , we will consider this case . constructing the trial functions
, one can suppose that they should be similar to the modes with @xmath173 .
eigenfunctions for the translational `` zero mode '' , @xmath174 and @xmath175 , are well known , see eq .
( [ eq : m=+/-1 ] ) . to construct the trial functions it is convenient to use functions @xmath176 and @xmath177 in the form @xmath178 which coincide with the exact `` zero mode '' eigenfunctions for the mode with @xmath5 in the limiting case of the bp
soliton.@xcite now we can choose the trial functions @xmath85 and @xmath86 as follows , @xmath179 the variational problem ( [ eq : functional ] ) can be solved analytically in the main approximation for @xmath180 using the explicit bp solutions .
tedious calculations lead to the eigenfrequency for small @xmath0 , @xmath181 unfortunately , this method is not exact and leads to a value of the constant @xmath182 which is larger than the numerical value .
this is caused by the fact that the trial functions , which are constructed from zero modes , have the wrong asymptotical behaviour ( [ eq : theta_r<<1 ] ) at large distance , @xmath32 , @xmath183 @xmath184 , instead of ( [ eq : u&v - loc - infty ] ) , i.e. @xmath185 , @xmath186 .
to improve this result , one can change the asymptotical behaviour of the functions ( [ eq : u+v4smallr ] ) by exponential factors : @xmath187 in this case the calculations can be done numerically only ; the value of the constant @xmath188 . for the approximate description of the eigenfrequency of the mode with @xmath5 in the intermediate case of arbitrary @xmath0
, we use the trial function @xmath189 providing the asymptotically correct behaviour , ( [ eq : omega_m=-1 ] ) and ( [ eq : omega_m=-1_small_r ] ) , this approximation reproduces to the numerical results with an accuracy of about @xmath190 , see fig .
[ fig : omega ] .
let us describe the scattering of magnons by a soliton . note that without a soliton , free magnon modes have the form @xmath191 where @xmath78 is a `` radial wave number '' , @xmath78 and @xmath192 are connected by the dispersion law ( [ eq : dispersion_law ] ) , and @xmath193 are bessel functions .
the free modes like @xmath194 play the role of the partial cylinder waves of a plane spin wave @xmath195 to describe magnon solutions in the presence of a soliton , one should note that far from the soliton , @xmath125 , the `` coupling potential '' @xmath106 is exponentially small .
therefore there are two asymptotically independent evp ; the magnon amplitude @xmath196 has the same form ( [ eq : v - loc - infty ] ) as for the local modes , but the function @xmath194 shows an oscillating behaviour like @xmath197 where @xmath198 are neumann functions .
the quantity @xmath199 derives from the soliton
magnon scattering ; it can be interpreted as the scattering amplitude .
we use the one parameter shooting method to study numerically the problem of soliton magnon scattering , as described in ref . for vortices in the antiferromagnet .
choosing the shooting parameter @xmath124 , see eq .
( [ eq : u&v4x<<1 ] ) , we `` kill '' the growing exponent in ( [ eq : v - loc - infty ] ) ; as a result we have obtained a well pronounced exponential decay for @xmath196 ; the corresponding oscillating solutions for @xmath194 differ from the asymptotical values ( [ eq : u - oscillating ] ) only in the vicinity of the soliton core , see fig .
[ fig : u_scat ] . the scattering amplitude was found from these data by comparison with the asymptotics ( [ eq : u - oscillating ] ) .
the results are the following : for all modes the scattering amplitude @xmath200 tends to zero as @xmath201 . in the long wavelength limit
the maximal scattering is related to the mode with @xmath5 , for which the behaviour of @xmath202 looks singular . however , other properties depend strongly on the soliton radius . in the case of large @xmath0 ,
the scattering amplitude @xmath200 is positive for all modes in the long wavelength limit .
there is a maximum of @xmath200 at about @xmath203 for modes with _ positive _ @xmath2 ; but there are two poles in the scattering amplitude for modes with _ negative _ @xmath2 , and one pole for @xmath9 .
naturally , there is no real divergence at such a pole : the physically observed phase of the scattering , or the phase shift @xmath204 , varies monotonically , see fig .
[ fig : delta4large_r ] .
thus , there are principally different pictures for positive and negative @xmath2 : the total phase shift is [ eq : delta^tot ] @xmath205 the scattering picture is quite different for the case of small radius solitons . each mode with @xmath142 has a single pole only , the position @xmath206 of which increases when @xmath2 grows .
for the special case of the mode with @xmath136 , the scattering amplitude is anomalously small ; for the mode with @xmath5 there appears a second pole with anomalously small values of @xmath207 .
these results are presented in fig .
[ fig : delta4small_r ] .
the total phase shift is @xmath208 there appears discontinuous change of the total phase shift from ( [ eq : delta^tot4larger ] ) to ( [ eq : delta^tot4smallr ] ) for all modes with @xmath209 for the cases of intermediate soliton radii .
this transition occurs independently for each mode at some critical values @xmath210 of the wave number .
for the analytical description of the scattering problem , let us analyze the evp qualitatively , replacing the real `` potentials '' ( [ eq : potentials ] ) in the evp ( [ eq : u&v ] ) by simplified ones , which are amenable to an exact treatment .
such an approach is well known in the quantum physics of solids as the pseudopotential method .
sometimes it is useful even when the potential is not small , see ref . for details .
let us discuss the choice of the pseudopotential .
the simplest way is to describe the soliton structure in the main approximation , using the ansatz @xmath211 which is a good approximation for both small and large distances .
replacing the soliton by this configuration , which does not have any in plane structure ( @xmath212 ) , reduces the complex evp ( [ eq : u&v ] ) to a very simple `` centrifugal model '' with @xmath213 : @xmath214 this model describes quasi free magnons in each of the regions @xmath215 and @xmath216 .
the only effect of the soliton
magnon interaction is a shift of the mode indexes in comparison with the free magnons ( [ eq : modes_free ] ) : @xmath217 the usual matching condition for these solutions has the form @xmath218_r = 0,\ ] ] where @xmath219_r\equiv ( \dots)\bigr|_{r+0 } - ( \dots)\bigr|_{r-0}$ ] .
calculations lead to the scattering amplitude in the form : @xmath220 where @xmath221 , see fig .
[ fig : delta - centrifugal ] . the total phase shift in this model @xmath222
the simple `` centrifugal model '' with the results ( [ eq : sigma - centrifugal ] ) explains qualitatively the scattering data for the small radius solitons , except for modes with @xmath224 , cf .
( [ eq : delta^tot4smallr ] ) .
the cause for these exceptions is the existence of truly local modes ( `` zero ones '' for @xmath134 , and another one with non zero frequency for @xmath5 ) ; while such modes are absent for the other @xmath2 .
the existence of a local mode diminishes the total phase shift by @xmath225 ( and decreases the number of states in the continuum spectrum by @xmath226 , due to the conserved total number of magnon states ) .
such a result is well known in 1d , see e.g. ref . . except for these cases
, one can expect for the scattering peculiarities at @xmath227 . in this range the approximation by an isotropic magnet and bp solitons is valid .
the scattering problem for the isotropic magnets was studied in detail in ref . .
over a wide range of wave numbers , @xmath228 , the asymptotics for the belavin
polyakov soliton are valid,@xcite [ eq : sigma4bp ] @xmath229 for larger wave numbers , @xmath230 , the scattering amplitude has a general , so
called , eikonal dependence , @xmath231 , @xcite @xmath232 our numerical data in this region are very close to the asymptotical behaviour ( [ eq : sigma4bp ] ) , see fig . [
fig : delta4small_r ] .
note that the maximal scattering in the region @xmath228 occurs for the rotational mode with @xmath9 , whose asymptotics ( [ eq : sigma4bp(m=0 ) ] ) has a singularity for @xmath201 .
there is one pole only for each mode ; its position corresponds to the pole in the belavin
polyakov solution , @xmath233.@xcite let us consider the local modes , for which the total phase shift is diminished by @xmath225 in comparison to the `` centrifugal model '' ( [ eq : delta^tot4centrifugal ] ) ; it is equal to @xmath234 , @xmath235 , @xmath236 for the cases @xmath136 , @xmath234 , @xmath237 , respectively .
there is one pole only for the simplest case of the mode with @xmath9 .
the scattering amplitude in the range @xmath238 is described by the formula ( [ eq : sigma4bp(m=0 ) ] ) .
the position of the pole is determined by the bp
result @xmath239.@xcite the reason for such a nice agreement with the model of the isotropic magnet will be explained below in the text .
there is no pole for the translational mode with @xmath136 .
this is due to a unique property of the isotropic model , where such a mode does not scatter at all;@xcite the magnon amplitude in this limit has the form @xmath240 which agrees with our numerical data , see fig .
[ fig : bp - amplitudes ] .
this unique property of reflectionlessness is caused by the high internal symmetry of the problem .
[ fig : delta4small_r](a ) shows that such a property is approximately valid for the small radius soliton in the anisotropic magnet , too : the scattering amplitude is anomalously small , so the total phase shift is absent .
another effect of this symmetry in the isotropic magnet is that the scattering problem for the bp model can be reduced to a simplified one with effectively smaller scattering .
such a `` reduced model '' has no local ( and quasi
local ) modes ; it is described by shifted indexes , @xmath241.@xcite it is natural to use the `` centrifugal '' approximation for such a reduced model ( [ eq : sigma - centrifugal ] ) . as a result ,
@xmath242 in the framework of the same `` centrifugal model '' ; for the other modes , eq .
( [ eq : delta^tot4centrifugal ] ) leads to the total phase shift @xmath243 . taking into account that @xmath244 , one can see that such a simple dependence for @xmath245 describes all modes , except of @xmath5 .
its behaviour is most interesting .
for the mode with @xmath5 there are two poles , while it should be one pole only for the belavin polyakov solitons .
indeed one of the poles lies in the region where the belavin
polyakov approximation is valid ; its value , @xmath246 , coincides exactly with the belavin
polyakov data.@xcite the second pole lies in the region @xmath247 , where the approximation by the isotropic magnet is not applicable .
this pole is caused by the presence of the local mode , the typical wave number of such a mechanism is @xmath248 .
the use of `` reduced '' or `` centrifugal '' models is certainly wrong for this case .
we will discuss the scattering amplitude in this region later , on the basis of a more general pseudopotential model , see eq .
( [ eq : sigma4m=-1(kr<<1 ) ] ) .
let us consider the case of large radius solitons , @xmath44 . here exist local modes for each @xmath2 . therefore the simple `` centrifugal model ''
as well as other simple models without a discrete spectrum seem to be wrong .
this can be explained by the fact that in the very simple model ( [ eq : centrifuge ] ) we omit the `` coupling potential '' ; the real potential @xmath98 tends to zero at the origin and at infinity , but has maxima at some finite distances .
thus we need a more general form of the pseudopotentials which accounts for the potentials @xmath250 , @xmath251 , and @xmath98 .
to describe the soliton shape , we can again use the expression ( [ eq : delta - ansatz ] ) for the out of plane soliton configuration .
however , we need to consider the in plane signature , too . according to ( [ eq : delta - ansatz ] )
, it could be described as @xmath252 therefore we can suppose that the `` coupling potential '' @xmath106 is nonzero only at some distance around the soliton radius and approximately replace it by a @xmath253potential as follows , [ eq : delta - potentials ] @xmath254 the same assumptions lead to simple expressions for the other potentials in the evp ( [ eq : u&v ] ) , @xmath255 here @xmath256 and @xmath257 are trial parameters .
we choose this pseudopotential model ( [ eq : delta - potentials ] ) out of the following reasons : ( i ) it allows an exact analytical solution ; ( ii ) it guarantees the correct asymptotic behaviour of the solutions ; ( iii ) it offers a possibility to identify the parameters of the pseudopotential from the discrete spectrum ( [ eq : omega_m_cases ] ) .
let us consider the discrete spectrum in the framework of the pseudopotential ( [ eq : delta - potentials ] ) . comparison with the eigenfrequencies ( [ eq : omega_m_cases ] ) provides a possibility to determine all trial parameters @xmath258 . for the discrete part of the spectrum , the evp ( [ eq : u&v ] ) with potentials ( [ eq : delta - potentials ] ) has solutions like [ eq : u&v - delta - pot ] @xmath259 now it is possible to calculate the eigenspectrum by matching the solutions ( [ eq : u - delta - pot ] ) and ( [ eq : v - delta - pot ] ) at the distance @xmath0 , [ eq : u&v - joining ] @xmath260_r = \frac{c_2}{\delta_0}\cdot u\bigr|_r- \frac{c_1}{\delta_0}\cdot v\bigr|_r , \\
% \label{eq : u&v - joining(2 ) } % & & \bigl[v^\prime\bigr]_r = \frac{c_2}{\delta_0}\cdot v\bigr|_r-\frac{c_1}{\delta_0}\cdot u\bigr|_r.\end{aligned}\ ] ] calculations lead to the eigenfrequencies in the form ( [ eq : omega_m ] ) . using the condition for the `` zero modes '' ( [ eq : omega=0 ] ) , the eigenfrequencies can be represented in the form ( [ eq : omega_m * ] ) ; the constants @xmath256 , @xmath257 can be identified by comparison with the spectrum ( [ eq : omega_m * ] ) , @xmath261 the parameters @xmath84 and @xmath162 were calculated by the variational approach , @xmath262 , @xmath263 , see eq .
( [ eq : alpha&beta ] ) .
let us discuss the scattering problem . solving the evp ( [ eq : u&v ] ) with pseudopotentials ( [ eq : delta - potentials ] )
, we obtain for the continuum spectrum , @xmath108 , the following solutions .
the function @xmath85 for @xmath215 has the form @xmath264 where @xmath265 . at large distances , @xmath216
, it has the usual oscillating form ( [ eq : u - oscillating ] ) .
the function @xmath86 has the localized form ( [ eq : v - delta - pot ] ) .
one can find the scattering amplitude by matching the solutions ( [ eq : u - oscillating ] ) , ( [ eq : u - delta - pot - scat ] ) , and ( [ eq : v - delta - pot ] ) at the distance @xmath0 , using the condition ( [ eq : u&v - joining ] ) : @xmath266 j_{|m+1|}(kr)- k\delta_0 j_{|m+1|}^{\prime}(kr)}{k\delta_0 y_{|m+1|}^{\prime}(kr)-\left[{\frak c}_m(k)+{\frak f}_m(k)\right ] y_{|m+1|}(kr)},\ ] ] here @xmath267 it is easy to analyze the expression for the scattering amplitude ( [ eq : sigma - pseudo ] ) in some limiting cases . in the long wavelength limit , @xmath268 , one can use the asymptotics for the cylindrical functions @xmath269 , @xmath270 at @xmath271 and , at the same time , the asymptotics for @xmath272 , @xmath273 for @xmath130 .
simple calculations show that the scattering intensity is maximal for the mode with @xmath5 , [ eq : sigma4kr<=1 ] @xmath274 which agrees with the numerical data , see fig .
[ fig : delta4large_r](b ) .
note , that the eq .
( [ eq : sigma4m=-1(kr<<1 ) ] ) was obtained using one condition only , @xmath275 .
therefore it is valid for the case of small @xmath0 , too , explaining the existence of the second pole for the mode @xmath5 both for large and small radii .
for the other values of @xmath2 , we can restore the general dependence of the amplitude of the magnon scattering on the large radius soliton , @xmath44 : @xmath276 ) , is positive for all modes in the long wavelength limit .
in the range @xmath277 the asymptotics for @xmath269 , @xmath270 for @xmath130 are valid , which in the main approximation for @xmath148 leads to a linear dependence of the scattering phase shift on @xmath278 , @xmath279 for this range of @xmath78 , all curves look like a family of parallel lines with equal distance of about @xmath280 .
note that the real values of the trial parameters @xmath256 and @xmath257 lead to the corrections in eqs .
( [ eq : sigma4kr<=1 ] ) like @xmath148 .
naturally the pseudopotential method can not explain the scattering results for @xmath281 , where the influence of truly local modes and the differences between the real potential and the @xmath253function are dominant .
nevertheless , the method gives quite good results even for the case @xmath282 , where the linear asymptotics does not work .
corresponding curves are plotted in fig .
[ fig : sigma - pseudo ] using the general dependence ( [ eq : sigma - pseudo ] ) with trial parameters @xmath256 and @xmath257 .
fitting of these parameters gives a result in the zeroth
order approach for @xmath148 : @xmath283 , @xmath284 instead of @xmath283 , @xmath285 , according to ( [ eq : c1-c2 ] ) .
the difference is caused by the limitations of the model ( [ eq : delta - potentials ] ) .
for example , the above mentioned model leads to an increase of the scattering phase when @xmath78 increases ; the total phase shift for this model is @xmath286 , as for the `` centrifugal model '' , see ( [ eq : delta^tot4centrifugal ] ) , i.e. the qualitative properties of these models are equivalent .
the appearance of a local mode diminishes the total phase shift by @xmath225 . as a result
, there is no total phase shift for all modes with positive @xmath2 ; but @xmath287 for all modes with @xmath288 .
the scattering picture in the short wavelength limit , @xmath230 , gives the general , so
called eikonal dependence @xmath289 . however , in contrast to the case of the isotropic magnet ( [ eq : sigma4bp(kr>>1 ) ] ) , the scattering amplitude has the same sign @xmath290 for all modes due to the presence of the effective attractive potential , which causes the existence of local modes .
when the soliton radius decreases and the local mode disappears , the sign of @xmath291 changes ; the result becomes similar to ( [ eq : sigma4bp(kr>>1 ) ] ) .
thus , we explain the particularities in the scattering picture of ea magnets , in comparison with the isotropic case , by the influence of local modes . to examine this influence ,
let us consider the scattering data @xmath291 for the modes @xmath292 and @xmath293 with different soliton radii @xmath0 as functions of the parameter @xmath278 , i.e. @xmath294 ; this gives a possibility to compare the scattering data with the isotropic model , see fig .
[ fig : delta_m_2.dif - r ] . for the small radii
the scattering picture is similar to the bp case ( solid lines in fig . [ fig : delta_m_2.dif - r ] ) . when the soliton radius increases , the scattering phase jumps up by @xmath225 , decreasing the total phase shift .
such a bifurcation takes place for the mode with @xmath292 at @xmath295 , and for the mode with @xmath5 at @xmath296 .
these values agree with the crossover points @xmath297 , see table [ table : rc ] .
thus , it is natural to suppose that the bifurcation is connected with the appearance of the quasi
local modes inside the continuum spectrum for @xmath298 .
simple calculations show that the positions @xmath207 of the poles correspond to the frequencies of quasi local modes , which can be estimated by eq .
( [ eq : omega_m(r>>1 ) ] ) .
the foregoing analysis of the spectrum for the modes with @xmath172 can be used to describe the translational motion of the soliton .
such an approach was first proposed in ref . , where an effective mass of the vortex in a 2d ep magnet was calculated from numerical data for the eigenmode spectrum of the finite size magnet .
an important progress for the vortices in fm was achieved in ref . ; the specific structure of the spectrum of modes with @xmath172 ( namely the existence of a goldstone mode with @xmath299 , and a number of doublets with mean frequency @xmath300 and a small splitting ) leads to non newtonian equations of motion with @xmath301 , @xmath302 , time derivatives .
such equations were derived and discussed earlier in ref . , using a phenomenological approach , for a review see ref . .
the analytical calculation@xcite verified their adequacy ; besides , it allows to calculate the coefficients in the equations of motion with an error of only about @xmath303 .
thus , it is possible to describe in this way such unusual properties as the appearance of the non newtonian equations of motion .
the spectrum of magnon modes with @xmath172 has a different structure for the isotropic fm .
first of all , there are no doublets;@xcite for an approximate description of the dynamics only two modes are sufficient : the translational goldstone mode with dependence @xmath299 and another mode with @xmath5 .
therefore , the effective equations of motion have a newtonian form with an effective mass , which diverges as @xmath304 , see ref . .
such behaviour agrees with the direct calculations done by zaspel.@xcite let us note that all magnets mentioned above , i.e. ep and isotropic systems , have a gapless dispersion law and , as a result , a strong interaction of the soliton with the boundaries of the magnet ; the interaction force is proportional to @xmath305 for the ep vortices@xcite and @xmath306 for the belavin
polyakov solitons.@xcite the absence of a gap manifests itself in the particularities of the dynamical coefficients ( the effective mass @xmath307 for the bp soliton in the isotropic fm@xcite and the coefficient @xmath308 in the term with @xmath309 $ ] for the vortex in the ep fm@xcite ) .
this can be explained as follows .
the existence of the mass @xmath307 and the gyroscopical force @xmath310 $ ] leads to a finite frequency of the larmor precession of the soliton . for a gapless dispersion law
, this frequency lies in the continuous spectrum .
thus , the larmor motion of the vortex leads to the generation of magnons . for a finite size system without dissipation ,
such magnons are distributed through the whole magnet ; the larmor dynamics of the soliton centre is strongly coupled with a `` magnon cloud '' , having a scale of about the system size .
therefore , the position of the soliton centre , which can be determined as a point with @xmath311 ( bp soliton ) , or @xmath312 ( vortex ) , in fact plays the role of a collective variable which governs the motion of the `` magnon cloud '' .
then it is not surprising that the corresponding equations of motion are nonlocal , leading to the divergence of the dynamic coefficients as @xmath313 .
the situation is quite different for the case of the ea fm .
our numerical analysis and analytical calculations show that the picture of doublets for local modes with @xmath314 and @xmath315 is valid for large @xmath0 , see sec .
[ sec : local ] . on the other hand ,
the eigenfrequencies of these modes lie in the gap for the large soliton radii , @xmath316 , so one can expect the existence of a finite mass in the effective equations of the soliton motion .
therefore for the phenomenological description of the soliton dynamics we can use the @xmath317order differential equation @xmath318 = \boldsymbol{f}_e \left(\boldsymbol{x}\right),\ ] ] which corresponds to the picture with the lowest doublet of local modes . note that further generalizations for the finite size magnet take into account the next doublets of quasi continuous modes . here
the hierarchy of the effective equations of motion containing only even
order time derivatives manifests itself . in eq .
( [ eq : eqmot ] ) @xmath319 describes the position of the soliton , @xmath307 is the mass coefficient , @xmath320 the gyrocoefficient , @xmath321 the unit vector along the easy ( @xmath322 ) axis , and @xmath323 the external force acting on the soliton due to the boundary or other solitons . assuming that deviations of the soliton from the equilibrium position are small , @xmath324 , the effective equation of motion ( [ eq : eqmot ] ) can be solved by a harmonic ansatz , which leads to @xmath325 for the localized solitons , the interaction between soliton and boundary ( or between two solitons ) has an exponential decay as @xmath326 , @xmath327 , see eq .
( [ eq : theta_r<<1 ] ) .
thus @xmath328 can be neglected for a large system size @xmath329 , so the frequencies ( [ eq : omega12+fe ] ) have the form : @xmath330 the zero frequency @xmath331 corresponds to the goldstone mode ( position shift of the soliton ) , in which all spins rotate with the frequency of the soliton precession @xmath30 in the laboratory frame .
the presence of the rotating frame frequencies in the phenomenological description of the soliton dynamics has a simple explanation : the frequencies ( [ eq : omega12 ] ) describe , in essence , the motion of the soliton centre .
let us discuss their link with the frequencies of small oscillations , see ( [ eq : vartheta&mu - via - u&v ] ) .
it is easy to write down expressions for the magnetization components in the laboratory coordinate frames : @xmath332\cdot e^{iq\chi+i\omega t},\\ % m_z & = & \cos\theta_0 - \sin\theta_0 \sum_{\alpha } \left ( u_\alpha + v_\alpha \right)\cos(m\chi+\tilde{\omega}_\alpha t).\end{aligned}\ ] ] one can see that the in plane components have no simple frequency dependence , including the combined values @xmath333 . at the same time , the out of plane components , @xmath334 , depend on the frequency @xmath80 in the rotating frame only . in our phenomenological description only the @xmath322components contain information about the position of the soliton centre @xmath319 , determined by the condition @xmath311 .
thus , just the frequency @xmath80 , which is calculated above and plotted in all figures , determines the soliton motion .
the solution of the effective equation ( [ eq : eqmot ] ) with the nonzero frequency @xmath335 allows us to calculate the effective mass of the soliton , @xmath336 .
the value of the gyroconstant is well known , @xmath337 , see refs . .
as a result , the soliton mass is finite for every soliton radius ( the local mode with @xmath335 exists for every @xmath0 , see fig .
[ fig : omega ] ) .
this fact corresponds qualitatively to the calculations in ref . .
the soliton mass we have obtained numerically from the @xmath335data is plotted in fig .
[ fig : mass ] as a function of @xmath0 . in the limit of a large radius soliton
the formula ( [ eq : omega_m=-1 ] ) yields the dependence [ eq : mass ] @xmath338 which agrees with the results obtained in ref . and our numerical data ,
see fig .
[ fig : mass ] . in the case of small radius , @xmath3 ,
the dependence ( [ eq : omega_m=-1_small_r ] ) is valid , thus @xmath339 where @xmath340 was introduced in eq .
( [ eq : theta_r<<1 ] ) .
this result corresponds to the one for the belavin
polyakov soliton,@xcite @xmath341 , because here the characteristic length @xmath340 is to be replaced by the system size @xmath329 .
we have studied the magnon soliton system in the model of the 2d heisenberg easy axis ferromagnet . combining numerical and analytical methods ,
we have obtained complete results about bound and scattering magnon states in the system . a rich spectrum of truly local modes was found along with bifurcations of these modes with the change of the soliton radius
. for the modes with higher azimuthal numbers @xmath2 , we have verified the picture of doublets with small splitting for large soliton radii @xmath0 .
the spectrum changes strongly with decreasing of @xmath0 : all modes with @xmath142 `` leave '' the region of the discrete spectrum , transforming to quasi local modes . such modes could be observed experimentally by soliton magnetic resonance as was done for 1d solitons.@xcite the influence of truly local modes is important for the scattering problem , leading to the bifurcations of the phase shift . as a result , when @xmath342 , the scattering picture is qualitatively the same as for the isotropic magnet which could be explained in the framework of a simple `` centrifugal model '' .
there is one exception only , @xmath5 , where the local mode exists for every soliton radius .
our investigations can be applied to the description of the eigenmodes of magnetic dots . in the present paper
we develop the theory of local modes on the soliton background ; it could be a good guide for the study of the normal modes in the vortex state magnetic dots .
our theory is constructed for the soliton in the model ( [ eq : l ] ) , where the soliton is stabilized by the internal precession .
it is clear that such a model can not guarantee the quantitative correspondence with the case of vortex state magnetic dots , where the static soliton structure is stabilized by the magnetic dipole interaction .
we did not consider this type of interaction in the paper , as it is difficult to account for .
nevertheless we believe that the main features of the problem studied above are generic .
first of all , we expect the appearance of modes with anomalously small frequencies , e.g. the mode of the translational oscillations of the vortex centre .
the non zero frequency of this mode is caused by the interaction with the boundary only .
secondly , doublets with @xmath343 , @xmath344 , should appear , when the radius of the vortex is rather large .
we have used the results on the local modes for the most interesting case of the translational modes with @xmath172 to describe the soliton motion in the infinite size magnet : it is possible to identify the soliton mass , which is finite due to the localized soliton structure .
in contrast to both the belavin polyakov soliton in the isotropic fm and the magnetic vortex in the easy plane fm , the soliton motion in our case of the easy axis fm is similar to the motion of a finite mass charged particle in a magnetic field .
d.d.sh thanks the university of bayreuth , where part of this work was performed as a post doc of the graduiertenkolleg `` nichtlineare spektroskopie und dynamik '' , for kind hospitality .
work at kiev was partially supported by intas97 - 31311 grant .
b. hillebrands , c. mathieu , c. hartmann , m. bauer , o. buettner , s. riedling , b. roos , s. o. demokritov , b. bartenlian , c. chappert , d. decanini , f. rousseaux , e. cambrill , a. mller , b. hoffmann , and u. hartmann , j. magn .
mater . * 175 * , 10 ( 1986 ) .
|
we study the magnon modes in the presence of a topological soliton in a 2d heisenberg easy axis ferromagnet .
the problem of magnon scattering on the soliton with arbitrary relation between the soliton radius @xmath0 and the `` magnetic length '' @xmath1 is investigated for partial modes with different values of the azimuthal quantum numbers @xmath2 .
truly local modes are shown to be present for all values of @xmath2 , when the soliton radius is enough large .
the eigenfrequencies of such internal modes are calculated analytically on limiting case of a large soliton radius and numerically for arbitrary soliton radius .
it is demonstrated that the model of an isotropic magnet , which admits an exact analytical investigation , is not adequate even for the limit of small radius solitons , @xmath3 : there exists a local mode with nonzero frequency .
we use the data about local modes to derive the effective equation of soliton motion ; this equation has the usual newtonian form in contrast to the case of the easy plane ferromagnet .
the effective mass of the soliton is found .
|
the general theory of qcd has been developed for more than 40 years , and at present , nobody ever doubts its validity .
however , on the other side there is still not a reliable way to deal with the long - distance effects of qcd which are responsible for the quark confinement and hadronic transition matrix elements , because their evaluations can not be done in perturbative approach .
thus , one needs to factorize the perturbative sub - processes and the non - perturbative parts which correspond to different energy scales .
the perturbative parts are in principle , calculable to any order within the framework of quantum field theory , whereas the non - perturbative part must be evaluated by either fitting data while its universality is assumed , or invoking concrete models .
the perturbative qcd method ( pqcd ) has been applied to study processes where transitions from heavy mesons or baryons to light hadrons are concerned @xcite , namely the pqcd which includes the sudakov resummation , is proved to be successful for handling processes with small 4-momentum transfer @xmath1 . indeed the processes involving heavy hadrons may provide us with an opportunity to study strong interaction , because compared to @xmath11 there exist natural energy scales ( heavy quark masses ) which can be used to factorize the perturbative contributions from the non - perturbative effects . on the other hand , for the processes involving heavy hadrons , at small recoil region , where @xmath12 is close to unity ( @xmath6 and @xmath7 denote the four - velocities of the initial and final hadrons ) , i.e. the momentum transfer @xmath1 is sufficiently large , the heavy quark effective theory ( hqet ) works well due to an extra symmetry @xmath13 @xcite .
therefore the hqet and pqcd seem to apply at different regions of @xmath1 . for a two - body decay ,
the momentum transfer is fixed by the kinematics , however , for a three - body decay , @xmath1 would span the two different regions . among all the processes , the semi - leptonic decay of hadrons plays an important role for probing the underlying principles and employed models because this process is relatively simple and less dependent on the non - perturbative qcd effects .
namely leptons do not participate in strong interaction , and there is no contamination from the crossed gluon - exchanges between quarks residing in different hadrons which are produced in the weak transitions , whereas such effects are important for the non - leptonic decays .
thus one might gain more model - independent information , such as extraction of the cabibbo - kobayashi - maskawa matrix elements from data . in the semi - leptonic decays of heavy hadrons
it is expected to factorize the perturbative and non - perturbative parts more naturally .
recently the delphi collaboration reported their measurement on the @xmath8 decay form factor in the semi - leptonic process @xmath14 and determined the parameter @xmath15 in the isgur - wise function @xmath16 @xcite .
there is a flood of papers to discuss the semi - leptonic decays of heavy mesons and the concerned factorization . by contraries ,
the studies on heavy baryons are much behind @xcite , because baryons consist of three constituent quarks and their inner structures are much more complicated than mesons . in this work
, we are going to employ the one - heavy - quark - one - light - diquark picture for the heavy @xmath8 and @xmath9 to evaluate the form factors of this semi - leptonic transition @xmath17 .
even though the subject of diquark is still in dispute , it is commonly believed that the quark - diquark picture may be a plausible description of baryons @xcite , especially for the heavy baryons which possess one or two heavy quarks .
the kinematic region for the semileptonic decay @xmath17 can be characterized by the quantity @xmath4 , which is defined as @xmath18 where @xmath19 are the four - momenta of @xmath8 and @xmath9 respectively .
it is noted that this parameter @xmath4 which is commonly adopted in the pqcd approach is exactly the same as the recoil factor @xmath5 used in the hqet .
the momentum transfer @xmath1 in the process is within the range of @xmath20 , equivalently , it is @xmath21 . in the framework of the hqet @xcite
, this process was investigated by some authors @xcite . for larger @xmath1 ,
the hqet works well , whereas one can expect that for smaller @xmath1 , the pqcd approach applies .
in this work , following ref.@xcite , we calculate the contribution from the region with small @xmath1 , i.e. @xmath22 , to the amplitude in pqcd .
one believes that the pqcd makes better sense in this region .
krner et al .
discussed similar cases and suggested that the symmetry for smaller recoil is different from that for larger recoil , so they used the isgur - wise function to obtain the amplitude in the kinematic region of smaller recoil , but brodsky - lepage function for larger recoils @xcite .
our strategy is similar that we apply the pqcd for small @xmath1 while apply hqet for large @xmath1 where the pqcd is no longer reliable , instead . concretely , when integrating the amplitude square from minimum to maximum of @xmath1 to gain the decay rate , we divide the whole kinematic region into two parts , small and large @xmath1 ( @xmath4 , equivalently ) .
we phenomenologically adopt a turning point at a certain @xmath4 value , to derive the form factors ( defined below in the text ) in terms of pqcd in the region from @xmath23 to this point and then beyond it we use the hqet instead @xcite .
we let the two parts connect at the turning point . from ref .
@xcite , we notice that as @xmath24 , the pqcd result is not reliable , so that we choose the turning point at vicinity of @xmath25 . to testify
if the choice is reasonable , we slightly vary the values of the turning point as choosing @xmath26 , @xmath27 and @xmath28 to see how sensitive the result is to the choice .
moreover , it is noted that as @xmath3 and 1.15 are chosen , the two parts connect almost smoothly .
even though , to make more sense , we adopt a proper numerical program to make the connection sufficiently smooth , namely we let not only the two parts connect , but also the derivatives from two sides are exactly equal . in fact , small differences in the derivatives are easily smeared out by the program .
later in the text , we will explicitly show that the final result is not much sensitive to it , thus one can trust its validity .
we name the scheme as the `` hybrid '' approach .
we also parameterize the form factor with respect to @xmath4 based on our numerical results .
in fact , when we integrate over the whole kinematic range of @xmath4 , we just use the parameterized expression .
moreover , we not only reevaluate the form factors @xmath29 and @xmath30 of the exclusive process in the diquark picture , but also calculate the form factors @xmath31 and @xmath32 which were neglected in previous works @xcite .
so far , the data on the @xmath8 semi - leptonic decay are only provided by the delphi collaboration @xcite and not rich enough to single out the contributions from @xmath31 and @xmath32 .
more accurate measurement in the future may offer information about them .
our treatment has another advantage . in the pure pqcd approach
, there is an end - point divergence at @xmath33 , even though it is mild and the decay rate which includes an integration over the phase space of final states , i.e. over @xmath4 , is finite . as calculating the contribution from the region with large @xmath1 (
@xmath34 is not zero , @xmath4 can not be exactly 1 , thus the superficial singularity does not exist at all , but the form factors are obviously proportional to @xmath35 which has the singular property . ] to the form factors in terms of the hqet , the end - point divergence does not exist at all .
we organize our paper as follows , in section ii , we derive the factorization formula for @xmath36 .
our numerical results are presented in section iii .
finally , section iv is devoted to some discussions and our conclusion .
the amplitude of @xmath37decay process is written as : @xmath38 where @xmath39 and @xmath40 are the momenta of @xmath41 and @xmath42 respectively .
according to its lorentz structure , the hadronic transition matrix element can be parameterized as @xmath43 \lambda_{b}(p),\end{aligned}\ ] ] where @xmath44 and @xmath45}/{2}$ ] . for the convenience of comparing with the works in literature , we rewrite the above equation in the following form according to ref.@xcite @xmath46 \lambda_{b}(p).\label{s2}\end{aligned}\ ] ] for the case of massless leptons , @xmath47 thus the form factors @xmath48 and @xmath49 result in null contributions .
the contributions from @xmath31 and @xmath32 were neglected in previous literature @xcite , nevertheless in our work , we will consider their contributions to the matrix elements and calculate them in terms of the diquark picture and our hybrid scheme .
the kinematic variables are defined as follows . in the rest frame of @xmath41 @xmath50 and @xmath51 the diquark momenta inside @xmath41 and @xmath42
are parameterized respectively as @xmath52 where @xmath53 , @xmath54 , @xmath55 and @xmath56 . according to the factorization theorem @xcite ,
the hadronic matrix element is factorized in the b - space as @xmath57 where @xmath58 .
the renormalization group evolution of the hard amplitude @xmath59 is shown as follows @xcite @xmath60 \widetilde{h}_{\mu}(x_1,x_2,{\mathbf{b}_{1}},{\mathbf{b}_{2}},r , m_{\lambda_b},t_{a(b)}),\end{aligned}\ ] ] where @xmath61 is the anomalous dimension .
the wave function of @xmath41 which has the heavy - quark and light - diquark structure , is given as @xcite @xmath62 where @xmath63 is the baryon spinor , and the superscript s denotes scalar diquark ( spin@xmath64 , isospin@xmath64 ) .
@xmath65 is a constant introduced in literature .
@xmath66 is the flavor component of the baryon , namely @xmath67}|0>$ ] , where @xmath68 and @xmath69}$ ] are the creation operator of b - quark and the scalar diquark of @xmath70quarks .
the @xmath42 distribution amplitude bears a similar form , @xmath71 including the sudakov evolution of hadronic wave functions , i.e. running the scale of wave function from @xmath72 down to @xmath73 @xcite : @xmath74\phi^{s}_{\lambda_b}(x_{1},{\bf{b}_1 } ) , \nonumber \\ & & \phi^{s}_{\lambda_c}(x_2,{\mathbf{b}_{2}},p',\mu)=\mathrm{exp}[-s(\omega_2,(1-x_2 ) p'^+)-2\int_{\omega_2}^{\mu } \frac{d \overline{\mu}}{\overline{\mu}}\gamma_q(\alpha_{s}(\overline{\mu } ) ) ] \phi^{s}_{\lambda_c}(x_{2},{\bf{b}_2})\label{suda},\end{aligned}\ ] ] where @xmath75(i=1,2 ) . in our work ,
the effective gluon - diquark vertices are defined as @xcite @xmath76 where @xmath77 .
ccc according to the factorization scheme which is depicted in fig.[t1 ] , it is straightforward to obtain the analytic expressions of the form factors @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 and @xmath83 , by comparing eqs .
( [ s2 ] ) with ( [ s3 ] ) . in the above derivation
, the following transformation has been used . @xmath84 and
the explicit expressions of a , b , c and d are given in the appendix .
then we can obtain the analytical form of @xmath85 and @xmath86 ( i=1,2,3 ) making use of the relations between them and @xmath87(j=1,2,3 ) listed below @xmath88 we do not display the expressions of @xmath48 and @xmath49 for the reason given above .
the form factors are integrations which convolute over three parts @xcite : the hard - part kernel function , the sudakov factor and the wave functions of the concerned hadrons as @xmath89 } , \nonumber\end{aligned}\ ] ] where the explicit expression of the kernel functions @xmath90 is given in the appendix for concision of the text .
the explicit form of the sudakov factor appearing in the above equations is given in ref.@xcite as @xmath91 where @xmath92 is the color factor .
the wave function @xmath93 is @xcite @xmath94^{3}},\end{aligned}\ ] ] and @xmath95},\end{aligned}\ ] ] where @xmath96 is the modified bessel function of the second kind .
if neglecting the transverse momentum @xmath97 , i.e. set @xmath98 , the wave function can be simplified as @xmath99^{2}}.\end{aligned}\ ] ] the normalization conditions are set as @xcite @xmath100 the first formula determines the normalization of the parton distribution of the baryon , whereas the second one is related to the effective mass of the light diquark @xmath101 , and the third formula reflects connection between hadronic matrix element of the kinematic operator @xmath102 and hadronic distribution amplitude . to satisfy the above three normalization conditions
, the parameters would take the following values @xmath103 gev , @xmath104 gev@xmath105 , @xmath106 , @xmath107 and @xmath108 , and in the following numerical evaluation , we will use them as inputs . for the wave function of @xmath42 ,
the expressions are the same as that for @xmath41 , while the corresponding parameters are @xmath109 gev , @xmath110gev@xmath105 , @xmath111 , @xmath112 and @xmath113 .
thus we can write the differential decay width as @xmath114,\end{aligned}\ ] ] with @xmath115
in the one - heavy - quark - one - light - diquark picture , the @xmath116 diquark in @xmath41 is considered as a scalar of color anti - triplet . to calculate the form factors in the framework of pqcd , one can adjust the product @xmath117 to fit the empirical formula @xmath118 given by the authors of ref.@xcite as about @xmath119 . nowadays ,
measurements have been done and with the data , we have made a new fit . ] . then the corresponding parameters are obtained as @xmath120 using these values
, we can continue to numerically estimate the form factors @xmath121 and @xmath122 .
fig.[t2](a ) shows that the form factor @xmath121 is exactly equal to @xmath123 in the heavy quark limit .
the form factor @xmath124 and @xmath125 are much smaller than @xmath121 and @xmath123 , thus they can in fact be safely neglected for the present experimental accuracy . in fig.[t2](b )
, we plot the dependence of the differential width @xmath126 on @xmath4 . although both @xmath121 and @xmath122 have an end - point divergence at @xmath127 in pqcd approach , the differential decay rate is finite .
if one extrapolates the pqcd calculation to the region with smaller @xmath4 values , we obtain the form factors @xmath128 and @xmath129 in that region where there are obvious end - point singularities at @xmath33 .
redo the computations with the extrapolation ( in the original work @xcite , the authors extend the tangent of the pqcd result at a small @xmath4 value to @xmath130 , so the end - point singularity is avoided ) and obtain @xmath131 which is about @xmath132 ( slightly smaller than the value of 2% guessed by the authors of ref.@xcite , because then no data were available . ) .
obviously , the calculation in pqcd depends on the factor @xmath117 , which regularly must be obtained by fitting the data of semileptonic decays , so that the theoretical predictions are less meaningful . instead
, we will use our hybrid scheme where we do not need to obtain the factor @xmath117 by fitting data , since the connection requirement substitutes the fitting procedure ( see below for details ) .
the transition rate was evaluated in terms of hqet by the authors of ref .
@xcite . according to the definitions given in eq.([gx ] ) , we re - calculate @xmath133 while dropping out @xmath134 and @xmath135 and also obtain similar conclusion that @xmath121 and @xmath122 are the same in amplitude , but opposite in sign , as shown in fig .
[ t2](a ) , whereas @xmath124 is very small and @xmath125 is exactly zero in hqet .
the theoretical prediction on the rate of the semi - leptonic decay @xmath17 in the hqet is @xmath136 and the branching ratio is @xmath137 .
the transition rate of @xmath37 has recently been measured by the delphi collaboration @xcite , and the value of @xmath138 is @xmath139 .
it is noted that the result calculated in terms of hqet is only consistent with data within two standard deviations . as widely discussed in literature , in the region with large @xmath1 ( small @xmath140values ) , the result of hqet is reliable , whereas for the region with small @xmath1 ( larger @xmath140values ) the pqcd is believed to work well .
therefore , to reconcile the two approaches which work in different @xmath4 regions , @xmath141 , we apply the hqet for small @xmath4 , but use pqcd for larger @xmath4 .
our strategy is that we let the form factors @xmath121 , @xmath122 derived in the pqcd approach be equal to the value obtained in terms of hqet at the point @xmath10 .
the numerical results of the form factor @xmath142 in the hybrid scheme are shown in fig.[t3 ] for three different @xmath143values : 1.05 , 1.10 and 1.15 , respectively .
it is noted that for @xmath3 and 1.15 , the left- and right - derivatives are very close and the connection is smooth , whereas , as @xmath10 is chosen as 1.05 , a difference between the derivatives at the two sides of @xmath144 obviously manifests .
we then adopt a proper numerical program to smoothen the curve , namely let the derivatives of the two sides meet each other for any @xmath140value near the cut point .
fig.[t3 ] shows that such treatment in fact does not change the general form of the curve , but makes it sufficiently smooth for all @xmath4 values including the selected cut point @xmath10 .
another advantage of adopting such a `` hybrid '' scheme is that there does not exist end - point singularity for the form factors at @xmath130 .
since at the turning point , we let the form factors derived in terms of pqcd be connected with that obtained in hqet , the product @xmath145 is automatically determined by the connection . with the value , we calculate the form factors within the range of small @xmath1 in pqcd . in this hybrid scheme , one does not need to invoke the data on the semi - leptonic decay to fix the parameter @xmath145 at all . by our numerical results obtained in the hybrid scheme
, the form factors @xmath121 ( or @xmath146 ) can be parameterized in a satisfactory expression , here we only present the expression for @xmath147 as @xmath148 and similar parameterized form factors were discussed in ref.@xcite .
the expression can be described by only one `` isger - wise function '' for the transition @xmath149 at the heavy quark limit , and it is done by the delphi collaboration based on their data on @xmath150 .
it is parameterized as @xcite @xmath151 where @xmath152 . obviously , this expression is only valid to the leading order , i.e. linearly proportional to @xmath153 where @xmath154 exactly corresponds to the parameter @xmath4 which is commonly adopted in the pqcd language .
by contrast , our result includes higher power terms because the @xmath155 corrections are automatically taken into account in our work .
it is noted that the coefficient of the linear term in our numerical result is reasonably consistent with the @xmath15 obtained by fitting data . to obtain the total decay width ,
we integrate over the whole range @xmath4 from 1 to @xmath23 , the integrand is the parameterized form factor in eq.([para ] ) .
we obtain @xmath156 for a comparison , we present the results corresponding to other two @xmath10 values where the smoothing treatment is employed , and they are @xmath157 one can notice that the factor @xmath145 does not change much as @xmath10 varies and they are about 1.5 times larger than the value obtained in pure pqcd ( eq.([pqcd ] ) ) .
when the turning point is chosen at @xmath144 , the branching ratio calculated in the hybrid scheme is very close to the result obtained in pure hqet , while for @xmath147 and @xmath158 , the resultant branching ratio is slightly larger than that obtained in pure hqet , but more coincides with the data . in our scenario ,
the hqet is applied for smaller @xmath4 , and the values of the form factors at @xmath10 are fixed by the theory .
the values can also determine @xmath145 which will be used for the pqcd calculations for larger @xmath4 .
the hqet is an ideal theoretical framework , but there is an unknown function which is fully governed by the non - perturbative qcd effects , that is the famous isgur - wise function
. the function can be either obtained by fitting data , or evaluated by concrete models .
various models would result in different slopes .
the authors of ref .
@xcite used the drell - yan type overlap integrals to obtain the slope which is what we employed to get the parametrization eq.([para ] ) and the slope is @xmath159 .
instead , the authors of @xcite evaluated the slope in the isgur - wise function by means of the qcd sum rules .
according to their result , we re - parameterize the form factor @xmath160 and have @xmath161 correspondingly , we obtain @xmath162 the fitted slope by the delphi collaboration is @xmath152 @xcite , which is between the two theoretical evaluated values . in these references ,
only linear term remained , due to uncertainties in the approximations the deviations are understandable .
therefore , one can note that there is a model - dependence which mainly manifests in the slope of the isgur - wise function . even though they deviate from each other at the linear term , the high power terms would compensate the deviation slightly and the predicted values on the branching ratio in two approaches eqs.([para],[para1 ] ) are qualitatively consistent with data .
there indeed is a byproduct which brings in an advantage that more accurate measurements can help to make judgement on validity of the models by which the isgur - wise functions are evaluated . to make more sense , we purposely present the ratio @xmath163 obtained in the hybrid scheme in fig .
[ t5 ] , it is noted that the ratio is qualitatively consistent with that given in ref .
@xcite which was shown on the left part of fig . 3 of their paper @xcite .
the authors of ref .
@xcite extended @xmath4 into the un - physical region ( @xmath164 for @xmath150 ) , while we only keep it within the physical region .
it is noted that in the physical region , numerically our result is very close to that obtained in ref .
but if one extends the curve to larger @xmath4 , he will notice that our curve is convex , but theirs is concave , namely the coefficient of the quadratic term has an opposite sign , but the difference is too tiny to be observed or bring up substantial difference for the evaluation of the decay width .
in this work , we investigate the semi - leptonic decay @xmath165 in the so called `` hybrid scheme '' and the diquark picture for heavy baryons @xmath8 and @xmath9 . the hybrid scheme means that for the range of smaller @xmath1 ( larger @xmath140value ) we use the pqcd approach , whereas the hqet for larger @xmath1 ( near @xmath130 ) , to calculate the form factors .
we find that the form factors @xmath124 and @xmath125 can be safely neglected as suggested in the literature .
besides , we do not need to determine the phenomenological parameters @xmath166 and @xmath167 by fitting data in the hybrid scheme as one did with the pure pqcd approach .
our result is generally consistent with the newly measured branching ratio @xmath0 within two standard deviations and the end - point singularities existing in the pqcd approach are completely avoided .
in fact , the final result somehow depends on the slope in the isgur - wise function of the hqet , which is obtained by model - dependent theoretical calculations .
therefore , we may only trust the obtained value to this accuracy , the further experimental data and development in theoretical framework will help to improve the accuracy of the theoretical predictions .
the quark - diquark picture seems to work well for dealing with the semi - leptonic decays of @xmath8 , and we may expect that the quark - diquark picture indeed reflects the physical reality and is applicable to the processes where baryons are involved , at least for the heavy baryons @xcite
. this picture will be further tested in the non - leptonic decays of heavy baryons .
we will employ the diquark picture and pqcd to further study the non - leptonic decay modes in our future work .
this work is partly supported by the national science foundation of china ( nsfc ) under contract no.10475042 , 10475085 and 10625525 .
we thank dr . c. liu for helpful discussions .
@xmath168hha+\big [ ( b c - a d ) r\ ( 1 + r ) - \ ( a^2 + a \nonumber\\ & & ( c - r ( d + 2 \rho - 2 b \rho ) ) + r ( b^2 r - 2 ( d r + c \rho ) - b ( c + ( -2 + d ) r\nonumber\\ & & + 2 c \rho ) ) ) \big]hhb\big\}{m^2_{\lambda_{b}}}f_{s}(\eta ) , \nonumber
\\ kernsgs_{g_{1}}&=&\big\{\big [ ( bc - ad ) ( r -1 ) + ( c^2 + a ( -2 + c + dr ) + ( b + d ) r ( dr - 2\rho ) -\nonumber\\ & & c ( r b - 2dr\rho - 2 rb\rho + 2 ) ) \big]hha+\big [ ( b c - a d ) \ r ( 1-r ) + ( a^2 + a \nonumber\\ & & ( c + r ( d -2 \rho + 2 b\rho ) ) + r ( b^2 r - 2 ( d r + c \rho ) + b\ ( ( -2 + d ) r -c\nonumber\\ & & + 2 c \rho ) ) ) \big]hhb\big\}\ { m^2_{\lambda_{b}}}f_{s}(\eta ) , \nonumber \\ kernsgs_{f_{2}}&= & ( a d - b c ) \ ( hha+ r\ hhb ) \
m^2_{\lambda_{b } } f_{s}(\eta ) , \nonumber \\ kernsgs_{g_{2}}&= & ( a d - b c ) \
( hha- r \
m^2_{\lambda_{b } } f_{s}(\eta ) , \nonumber \\\end{aligned}\ ] ] h. li and g. sterman , nucl . phys . * b 381*,129 ( 1992 ) ; g. korchemsky and g. sterman , phys . lett .
* b 340 * , 96 ( 1994 ) ; j. botts and g. sterman , nucl . phys .
* b 325 * , 62 ( 1989 )
. y. keum , h. li , a.i .
sanda , phys .
* b 504 * , 6 ( 2001 ) ; phys . rev . * d 63 * , 054008 ( 2001 ) ; c. l , k. ukai , m. yang , phys . rev . *
d 63 * , 074009 ( 2001 ) .
h. shih , s. lee and h. li , phys .
* d 61 * , 114002 ( 2000 ) .
n. isgur and m. wise , phys .
b232 * , 113 ( 1989 ) ; n. isgur and m. wise , nucl.phys .
* b 348 * , 276 ( 1991 ) ; h. georgi , _ ibid _ * b348 * , 293 ( 1991 ) .
j. abdallah et al , delphi collaboration , phys .
b585 * , 63 ( 2004 ) .
m. huang , h. jin , j. k@xmath176rner and c. liu phys .
b629 * , 27 ( 2005 ) .
y. dai , c. huang , m. huang and c. liu , phys . lett .
* b 387 * , 379 ( 1996 ) ; y. dai , c. huang , c. liu and c. d. l , _ ibid _ * b 371 * , 99 ( 1996 ) ; c. liu phys . rev . *
d 57 * , 1991 ( 1998 ) ; j. lee , c. liu and h. song , _ ibid _ * d 58 * , 014013 ( 1998 )
. x. h. guo , p. kroll , z. phys .
* c 59 * , 567 ( 1993 ) ; f. wilczek , hep - ph/0409168 .
a. flak and m. neubert , phys .
* d 47 * , 2982 ( 1993 ) ; b. holdom , m. sutherland and j. mureika , _ ibid _ * d49 * , 2359 ( 1994 ) . j. k@xmath176rner and p. kroll , z. phys .
* c 57 * , 383 ( 1993 ) . h. li , phys . rev . * d 52 * , 3958 ( 1995 ) .
h. li and h. yu , phys.rev.*d 53 * , 4970 ( 1996 ) .
h. shih , s. lee and h. li , phys.rev . *
d 59 * , 094014 ( 1999 ) ; c.h .
chou , h. shih , s. lee and h. li , phys .
d 65 * , 074030 ( 2002 ) .
r. jakob , p. kroll , m. schuermann and w. schweiger , z. phys . * a 347 * , 109 ( 1993 ) . c. peterson , d. schlatter , i. schmitt and p. zerwas , phys
d 27 * , 105 ( 1983 ) .
h. cheng , c. cheung and c. hwang , phys . rev .
* d 69 * , 074025 ( 2004 ) .
x. guo and t. muta , phys
d 54 * , 4629 ( 1996 ) ; x. guo , a. thomas and a. williams , _ ibid _ * d 59 * , 116007 ( 1999 ) .
|
in this work we use the heavy - quark - light - diquark picture to study the semileptonic decay @xmath0 in the so - called hybrid scheme .
namely , we apply the heavy quark effective theory ( hqet ) for larger @xmath1 ( corresponding to small recoil ) , which is the invariant mass square of @xmath2 , whereas the perturbative qcd approach for smaller @xmath1 to calculate the form factors .
the turning point where we require the form factors derived in the two approaches to be connected , is chosen near @xmath3 .
it is noted that the kinematic parameter @xmath4 which is usually adopted in the perturbative qcd approach , is in fact exactly the same as the recoil factor @xmath5 used in hqet where @xmath6 , @xmath7 are the four velocities of @xmath8 and @xmath9 respectively .
we find that the final result is not much sensitive to the choice , so that it is relatively reliable .
moreover , we apply a proper numerical program within a small range around @xmath10 to make the connection sufficiently smooth and we parameterize the form factor by fitting the curve gained in the hybrid scheme .
the expression and involved parameters can be compared with the ones gained by fitting the experimental data . in this scheme
the end - point singularities do not appear at all .
the calculated value is satisfactorily consistent with the data which is recently measured by the delphi collaboration within two standard deviations .
|
in recent years , experimental studies of transport properties of high - mobility two - dimensional ( 2d ) electron gas in weak perpendicular magnetic fields have uncovered a variety of remarkable quantum phenomena caused by transitions of electrons between different landau levels .
such transitions can lead to oscillations of dissipative resistance as a function of the magnetic field .
for example , under steady - state microwave illumination of the 2d gas the resistance oscillates with a period determined by the ratio of the radiation frequency to the cyclotron frequency @xmath0 .
this phenomenon is known as the microwave - induced resistance oscillations ( miro).@xmath1 for a sufficiently high radiation power the miro minima evolve into the intervals of magnetic field where the dissipative resistance vanishes.@xmath2 next , it was found that an increase in the electric current passing though the 2d layer substantially reduces the resistance@xmath3 and leads to oscillations of the resistance as a function of either the magnetic field or the current.@xmath4 such oscillations are controlled by the ratio of a characteristic energy , which is defined as the drop of the hall electric field across the classical cyclotron diameter , to the cyclotron energy .
this phenomenon has been called the hall field - induced resistance oscillations ( hiro ) .
it is also described in terms of zener tunneling between landau levels .
finally , in the systems with two occupied 2d subbands the resistivity oscillates as a function of the ratio of subband separation energy to the cyclotron energy owing to coupling of the subbands via intersubband scattering .
this phenomenon , called the magneto - intersubband oscillations ( miso ) , is important in the systems with small subband separation , such as the double quantum wells.@xmath5 all these kinds of oscillations are insensitive to positions of the landau levels with respect to the fermi energy .
therefore , unlike the shubnikov - de haas oscillations , they are not damped exponentially with increasing temperature .
the oscillatory phenomena described above require the presence of microwave illumination , strong hall field , or intersubband coupling to enable resonant transitions between different landau levels via elastic scattering of electrons by impurities or other static imperfections of the system .
the elastic scattering provides the main contribution to resistivity at low temperatures . in high - mobility samples
, however , inelastic scattering of electrons by acoustic phonons also contributes to transport , and can exceed the impurity - scattering contribution at the temperatures of several k. therefore , acoustic - phonon - assisted transitions of electrons between landau levels become important .
interestingly enough , these transitions in 2d case possess a resonant property and , for this reason , also lead to oscillations of the resistance .
oscillations of this origin have been discovered recently.@xmath6 an analysis of experimental data has suggested that the resistivity acquires an oscillating contribution whose periodicity is determined by the resonant condition @xmath7 where @xmath8 is the fermi momentum of electrons , @xmath9 is the constant of the order of sound velocity , and @xmath10 is an integer.@xmath6 this phenomenon has been called the phonon - induced resistance oscillations ( piro ) .
similar oscillations have been observed later in the phonon drag thermal power measurements.@xmath11 since the quantity in the left - hand side of eq .
( 1 ) is a characteristic phonon energy , the piro can be viewed as magnetophonon oscillations . for the case of electron scattering by optical phonons ,
the magnetophonon oscillations are known for a long time@xmath12 and have been observed both in bulk and 2d@xmath13 systems .
the special feature of piro , in contrast to the oscillations owing to scattering by optical phonons , is the dependence of magnetophonon resonance on the electron momentum . by emitting or absorbing acoustic phonons , the electrons jump between the landau levels , and the maximum probability of such transitions
is realized under electron backscattering condition , when the phonon wave number @xmath14 reaches its maximum value @xmath15 in the 2d plane .
initially,@xmath6 this resonant property of the electron - phonon scattering was explained by involving the interface phonon model .
later , by analyzing phonon - assisted transitions between landau levels , it was found@xmath11 that the bulk phonon model gives similar results .
it is worth noting that the maximum probability of the backscattering processes ( @xmath16 ) is not related to the presence of the magnetic field .
it reflects a fundamental property of the kinematics of two - dimensional electron scattering by three - dimensional acoustic modes with frequency @xmath17 , where @xmath18 and @xmath19 are the in - plane and out - of - plane components of the phonon wave vector .
it can be shown that the scattering probability , as a function of the phonon frequency , has a logarithmic singularity at the point of transition from the region @xmath20 , when pure 2d scattering ( @xmath21 ) is possible , to the region @xmath22 , when emission of three - dimensional phonons ( with finite @xmath19 ) becomes necessary .
the systematic studies of the piro are now under way.@xmath23 however , current understanding of this phenomenon is far from complete .
the characteristic phonon velocity @xmath9 entering eq .
( 1 ) varies in different experiments@xmath24 in the range from 2.9 to 5.9 km / s , and the origin of such variations is not clear .
the problem of the phase of the oscillations has not been discussed .
the dependence of the amplitude of the oscillations on temperature and magnetic field has not been investigated in detail , though general features of the experimental dependence have been successfully related@xmath25 to the behavior of the density of electron states in magnetic field . the main difficulty for interpretation of the experimental magnetoresistance is the complicated nature of acoustic - phonon modes . even in the case when the influence of interfaces on the phonon spectrum is neglected ( bulk phonon approximation ) , one should take into account three different phonon branches characterized by anisotropic ( i.e. , dependent on the direction of phonon momentum ) velocities . in this paper ,
the theory of piro is developed within the bulk phonon approximation .
both deformation - potential and piezoelectric mechanisms of electron - phonon interaction are included into consideration .
the anisotropy of phonon spectrum is taken into account .
the calculations are carried out for gaas quantum well layers grown in the [ 001 ] crystallographic direction .
apart from the numerical results , approximate analytical expressions for the oscillating part of the phonon - induced resistivity are presented .
the results are compared with available experimental data for two kinds of gaas layers with distinctly different densities and mobilities of 2d electrons .
the paper is organized as follows .
section ii contains general description of the resistivity of 2d electrons interacting with acoustic phonons in a weak magnetic field .
section iii includes both analytical and numerical results for the resistivity , their comparison with experiment , and discussion .
the summary and conclusions are given in the last section .
experimental investigations of phonon - induced resistance oscillations are carried out at low temperatutes @xmath26 satisfying the condition of degenerate electron gas , @xmath27 , where @xmath28 is the fermi energy .
magnetic fields are weak enough to have many landau levels populated , @xmath29 , so the electron transport possesses some quasiclassical features ( for example , the electron momentum is still a good quantum number ) . in the same time , the high quality of the samples ensures that the landau quantization is not suppressed by scattering at such low magnetic fields .
this means that the ratio of the cyclotron frequency @xmath30 to inverse quantum lifetime @xmath31 is not too small .
next , since in the high - mobility samples the ratio of transport time to quantum lifetime is large , the oscillations are observed in the regime of classically strong magnetic fields , when the cyclotron frequency is much larger than the inverse transport time @xmath32 .
the in - plane transport in this case can be viewed as hopping between cyclotron orbit centers , and the dissipative conductivity @xmath33 is represented in the form @xmath34 \nu_{\delta \delta^{\prime } } ( x_{\delta } - x_{\delta^{\prime}})^2/2 , % 2\ ] ] where @xmath35 is the index of a quantum state of electron , @xmath36 is the coordinate of the cyclotron orbit center in the direction of motion , @xmath37 is the equilibrium fermi - dirac distribution function , @xmath38 is the normalization area , and @xmath39 is the total scattering rate , which is represented as a sum of impurity and phonon scattering contributions : @xmath40 . assuming interaction of electrons with equilibrium phonons , the latter contribution is written as @xmath41 , % 3\end{aligned}\ ] ] where @xmath42 , @xmath43 , and @xmath44 are the phonon mode index , wave vector , and frequency , respectively .
next , @xmath45^{-1}$ ] is the planck distribution function .
the interaction is described by the function @xmath46 , which includes both deformation - potential and piezoelectric coupling of electrons to crystal vibrations . considering only acoustic modes in cubic crystals , the general form of this function
is written as follows : @xmath47 , % 4\end{aligned}\ ] ] where @xmath48 is the deformation potential constant , @xmath49 is the piezoelectric coupling constant , and @xmath50 is the material density .
the sums are taken over cartesian coordinate indices , @xmath51 are the components of the unit vector of the mode polarization , and the coefficient @xmath52 is equal to unity if all the indices @xmath53 are different and equal to zero otherwise .
the polarization vectors and the corresponding phonon mode frequencies are found from the eigenstate problem @xmath54 { \rm e}_{\lambda { \bf q } j}=0 .
% 5\ ] ] the dynamical matrix for cubic crystals , @xmath55 \delta_{ij } \nonumber \\ + ( c_{12}+c_{44 } ) q_i q_j(1-\delta_{ij } ) , % 6\end{aligned}\ ] ] is written in the elastic approximation and expressed through three elastic constants for which the conventional notations are used .
equation ( 5 ) has three solutions describing a high - energy mode ( la ) , which becomes purely longitudinal for high - symmetry directions ( [ 100],[110 ] , and [ 111 ] ) , and two low - energy modes ( ta ) which become purely transverse for these directions . very often , theoretical studies of transport properties of electrons in solids with cubic lattice symmetry are based on the continuum approximation for elastic vibrations , when @xmath56 and the phonon spectrum is given by three isotropic branches including one longitudinal mode and two degenerate transverse modes .
whereas this approach gives a reasonably good description of phonon - limited mobilities and applies for calculation of many kinetic coefficients , is not sufficient for description of phonon - induced resistance oscillations in magnetic field , because the anisotropy of the acoustic phonon branches is not weak in most semiconductors and becomes essential in evaluation of resonant scattering between landau levels .
the quantum states @xmath35 in eq .
( 2 ) can be treated as exact eigenstates of electrons interacting with randomly distributed impurities in the presence of a magnetic field .
it is convenient to rewrite eq .
( 2 ) through the green s functions of electrons . using the basis of landau eigenstates and specifying the growth axis of the quantum - well layer as @xmath57 ( i.e. , the @xmath58 $ ] crystallographic direction ) ,
one obtains the phonon - induced contribution to the conductivity : @xmath59 a_{\varepsilon}(n ' ) a_{\varepsilon - \hbar \omega_{\lambda { \bf q}}}(n ) , % 7\end{aligned}\ ] ] where @xmath18 and @xmath60 are the absolute value and polar angle of the in - plane component of phonon wave vector , @xmath10 are the landau level numbers , and @xmath61 is the magnetic length .
the function @xmath62 ^ 2 $ ] , where @xmath63 are the laguerre polynomials , describes scattering in the magnetic field .
the squared matrix element of a plane - wave factor , @xmath64 , is determined by the confinement potential which defines the ground state of 2d electrons , @xmath65 . applying the model of a deep rectangular quantum well of width @xmath66
, one can rewrite this squared matrix element as @xmath67 , where @xmath68^{-2}$ ] . finally , the spectral function @xmath69 , which is equal to the imaginary part of the single - electron ( advanced ) green s function divided by @xmath70 , characterizes disorder - induced broadening of electron states . in the born approximation ,
this broadening is conveniently described in terms of the quantum lifetime of electron , @xmath71 . in the absence of impurities ( the collisionless limit )
@xmath69 is reduced to the delta - function @xmath72 , where @xmath73 is the landau quantization energy .
the dissipative resistivity measured in experiments using hall bars is related to the conductivity @xmath33 according to @xmath74 , where @xmath75 is the classical hall conductivity and @xmath76 is the sheet carrier density . the resistivity is represented as a sum of contributions from electron - impurity and electron - phonon scattering : @xmath77 where @xmath78 and @xmath79 are the partial scattering rates which depend on the magnetic field . both @xmath80 and @xmath79
are straightforwardly obtained from @xmath81 of eq .
if the magnetic field is zero , @xmath78 and @xmath79 are reduced to the inverse transport times due to electron - impurity and electron - phonon interactions , respectively , and eq .
( 8) describes the classical ( drude ) resistivity .
let us consider the case of weak magnetic fields , when many landau levels are occupied
. then one can use smallness of the cyclotron energy @xmath82 , phonon energy @xmath83 , disorder - broadening energy @xmath84 , and temperature @xmath26 with respect to the fermi energy for evaluation of the expression ( 7 ) .
owing to the sharp form of the spectral functions , the contribution into the @xmath10-sums in eq .
( 7 ) comes from a limited number of landau levels in the vicinity of the fermi energy , and these sums can be approximately calculated with the following result : @xmath85 d({\varepsilon } ) d(\varepsilon - \hbar \omega_{\lambda { \bf q } } ) , % 9\end{aligned}\ ] ] where @xmath86 is the dimensionless ( i.e. , normalized to its zero - magnetic - field value ) density of electron states .
the integration over the in - plane phonon wave number @xmath18 is replaced in eq .
( 9 ) with the integration over the scattering angle @xmath87 according to @xmath88 .
this emphasizes that the phonon - assisted transport at large filling factors is described within the picture of quasielastic scattering of 2d electrons in the vicinity of the fermi surface . under the same assumptions ,
the contribution to the resistivity owing to electron - impurity scattering is given by @xmath89 where @xmath90 is the transport time for electron - impurity scattering .
it is defined according to @xmath91 , where @xmath92 is the fourier transform of the random impurity - potential correlator .
the self - consistent born approximation leads to the density of states in the form of an expansion in oscillation harmonics weighted with powers of the dingle factor @xmath93:@xmath94 @xmath95 if the magnetic fields is weak , the dingle factor is small .
this corresponds to the case of overlapping landau levels , when the density of states is given by a single - harmonic expression , @xmath96 .
in strong enough magnetic fields the electron system is in the regime of separated landau levels , when @xmath97 is represented by a periodic sequence of semi - elliptic peaks centered at the landau quantization energies @xmath98 .
the expressions ( 8)-(11 ) , together with eqs .
( 4)-(6 ) defining the phonon spectrum @xmath44 and the function @xmath46 , give a complete description of the phonon - assisted 2d magnetotransport at large filling factors .
the next step in evaluation of the resistivity can be done by calculating the integral over the energy @xmath99 . in this procedure , it is reasonable to omit the terms containing oscillating functions of energy under the integral .
such terms are responsible for the shubnikov - de haas oscillations , and they are exponentially suppressed with increasing temperature .
thus , the approximation used below corresponds to the condition @xmath100 which means that the temperature should not be too low . as a result , @xmath101 \equiv \nu_{ph}^{(0 ) } + 2 \sum_{k=1}^{\infty } a^2_k \nu_{ph}^{(k ) } , % 13\end{aligned}\ ] ] where @xmath102 ^ 2 $ ] and @xmath103 .
the quantity @xmath104 is an anisotropic scattering rate introduced for convenience purpose .
the corresponding result for the impurity - assisted magnetotransport has a simpler form , @xmath105/\tau^{im}_{tr}$ ] , which describes a positive magnetoresistance oving to landau quantization.@xmath106 the fermi momentum entering eqs .
( 13 ) and ( 14 ) is related to electron density according to @xmath107 .
the expression ( 13 ) is the central part of this paper .
the first term of this expression , @xmath108 , is responsible for phonon - induced contribution to background resistivity , while the second one corresponds to oscillations of the resistivity as a function of the magnetic field .
these oscillations ( piro ) are described in eq .
( 13 ) through a sum of oscillatory harmonics of the scattering rate , @xmath109 . because of the complicated form of phonon spectrum , the integrals in eq .
( 13 ) can not be calculated analytically in the general case .
however , an analytical consideration is possible for the oscillating part of @xmath79 , since this part contains rapidly varying functions of @xmath44 under the integral . to carry out the approximate integration
, it is convenient to represent expressions for oscillatory harmonics by using the spherical coordinate system according to @xmath110 , @xmath111 , where @xmath112 is the absolute value of phonon wave vector in the dimensionless form and @xmath113 is the inclination angle . in these variables , the phonon spectrum is conveniently written as @xmath114 , where @xmath115 is the anisotropic sound velocity for the mode @xmath42 .
the angles @xmath60 and @xmath113 determine the direction of the phonon wave vector . as a result , @xmath116 \cos \frac{4 \pi k p_f s^{(\lambda)}_{\varphi \chi } u}{\hbar \omega_c } , % 15\end{aligned}\ ] ] where @xmath117 given by eq .
( 14 ) is written as the expression in the square brackets ; the deformation - potential ( da ) and piezoelectric ( pa ) contributions are separated according to eq .
the main contribution to the integral over @xmath118 comes from the region @xmath119 , and the convergence of this integral is determined by the rapidly oscillating cosine function .
the other @xmath118-dependent functions , @xmath120 and @xmath121 , are slowly varying on the scale of convergence because of @xmath122 and because of condition ( 12 ) , respectively . after integrating over @xmath118 ,
one obtains @xmath123 \cos \left [ \frac{4 \pi k p_f s^{(\lambda)}_{\varphi \chi}}{\hbar \omega_c \sin \chi } -\frac{\pi}{4 } \right].\end{aligned}\ ] ] the remaining double integral over angular variables can be calculated using the method of fastest descent .
indeed , under the integral we have a rapidly oscillating function ( the argument of the cosine in eq .
( 16 ) is assumed to be large ) , and the main contribution comes from certain regions of @xmath60 and @xmath113 where @xmath124 varies most slowly .
naturally , these regions are in the close vicinity of the extrema points of the function @xmath125 .
the anisotropic sound velocity @xmath115 itself has a number of local maxima , minima , and saddle points , whose positions coincide with some high - symmetry directions in the reciprocal ( momentum ) space .
the number of extrema for @xmath125 is smaller because of the factor @xmath126 . for the mode with the highest velocity (
this mode becomes purely longitudinal for high - symmetry directions ) the anisotropy is not strong , and all the extrema are only at @xmath127 ( zero @xmath19 ) .
the slow modes ( which become purely transverse for high - symmetry directions ) also have extrema at @xmath127 .
since the anisotropy of these modes is stronger , they also may have additional groups of extrema at @xmath128 , where @xmath129 .
nevertheless , the main contribution of slow ( low - energy ) modes to oscillating magnetoresistance is associated with the extremum @xmath127 for a single mode whose velocity at @xmath127 is independent of the polar angle @xmath60 .
this mode at @xmath127 is purely transverse and polarized perpendicular to the quantum well plane ( direction [ 001 ] ) . in summary
, the analysis shows that the modes effectively contributing to the oscillating part of resistivity are : i ) the transverse mode polarized along [ 001 ] , whose velocity @xmath130 is independent of @xmath60 and corresponds to a maximum of @xmath125 as a function of @xmath113 ; ii ) the longitudinal mode with velocity @xmath131 polarized along [ 100 ] or along equivalent directions ( [ 010 ] , [ @xmath13200 ] , and [ 0@xmath1320 ] ) corresponding to local minima of @xmath125 ; iii ) the longitudinal mode with velocity @xmath133 polarized along [ 110 ] or along equivalent directions ( [ 1@xmath1320 ] , [ @xmath13210 ] , and [ @xmath1340 ] ) corresponding to saddle points of @xmath125 .
accordingly , the oscillatory harmonics of the scattering rate are written as sums of three components : @xmath135 where @xmath136 @xmath137 @xmath138 the coefficients of order unity standing in the denominators of these expressions are given by @xmath139 @xmath140 and @xmath141 where @xmath142 is the positive quantity characterizing the anisotropy of phonon spectrum . the contribution from transverse phonons
is associated with piezoelectric potential , while the contribution from longitudinal phonons is due to deformation potential of electron - phonon interaction . the accuracy of the analytical approach leading to eqs .
( 17)-(20 ) is not expected to be high for lowest - order piro peaks .
nevertheless , the applicability of these equations is steadily improved as one moves to lower magnetic fields and the argument of the oscillating function becomes larger .
for such low fields , the rate @xmath143 tends to overcome the rates @xmath144 and @xmath145 .
however , in the samples with higher electron density , where the characteristic phonon wave number @xmath15 is larger , the deformation - potential mechanism is more significant than the piezoelectric one
. therefore , both transverse - phonon and longitudinal - phonon contributions are important and should be taken into account .
notice also that for low magnetic fields corresponding to the case of overlapping landau levels one can take only the lowest harmonic @xmath146 in the sum in eq .
this reduces the oscillating part of phonon - induced transport rate @xmath79 to a simpler form @xmath147 . for completeness
, one can also present the result obtained in the isotropic approximation , where there are a longitudinal mode and two degenerate transverse modes with velocities @xmath148 and @xmath149 , respectively : @xmath150 in this approximation , the phase of the oscillations , as well as the dependence on magnetic field and fermi momentum is different from the case of anisotropic phonon spectrum .
k , the partial oscillating contributions from high - energy branch ( la ) and low - energy branches ( ta ) are also shown ( shifted for clarity).,width=347 ] the numerical calculations described below use the following elastic constants for gaas ( in units @xmath151 dyn/@xmath152 ) : @xmath153(k ) , @xmath154(k ) , and @xmath155(k ) , taken from a semiconductor material database.@xmath156 the temperature dependence of these constants is not really essential in the range of @xmath26 corresponding to experimental studies of oscillating magnetoresistance .
the other gaas parameters used are @xmath157 ev , @xmath158 v / nm , and @xmath159 g/@xmath160 .
an example of the results of numerical calculations according to eq .
( 13 ) is presented in fig .
1 . the calculation employs the following parameters of the high - mobility quantum well studied in ref .
17 : @xmath161 @xmath162 , @xmath163 nm , impurity - limited mobility @xmath164 @xmath152/v s , and quantum lifetime owing to impurity scattering @xmath165 ps .
the theoretical plots are in agreement with experimental data of ref .
17 as concerns the background resistance , piro peak positions , and the amplitude of the oscillations ( for example , at @xmath166 k the amplitudes of the peaks numbered 1 and 2 are 0.16 and 0.075 ohm per square , respectively ) .
the calculation shows that the phonon - induced resistivity is determined by both longitudinal and transverse modes , whose contributions are nearly equal to each other at the fields @xmath167 t and at 6 k. at lower fields the transverse - mode contribution becomes more important . with decreasing temperature both contributions
are suppressed according to the factor @xmath168 .
the la - mode contribution is suppressed stronger because the thermal activation energy for this mode is larger .
figure 1 shows that the peak placed slightly below 0.6 t is the lowest - order one for the la - mode contribution , so the origin of this peak ( discussed in ref .
17 ) seems to be clear .
however , the relative amplitude of this peak is higher than that observed in the experiment , which is possibly related to overestimation of the deformation - potential interaction in comparison to the piezoelectric one.@xmath169 the other low - order peaks ( @xmath170 , 2 , 3 ) are formed by both transverse - mode and longitudinal - mode contributions , so their positions @xmath171 are not expected to follow exactly a @xmath172-periodic dependence . nevertheless , the deviations appear to be small , and a 9-point linear fit of theoretical peak positions ( see fig .
2 ) gives the dependence @xmath173 , where @xmath174 km / s is very close both to the velocity @xmath175 km / s and to the velocity found experimentally,@xmath25 while @xmath176 is slightly larger than that obtained from the experiment.@xmath25 the parameter @xmath177 determines the phase of the magnetoresistance oscillations , which appears to be close to the phase of the oscillating factor in the approximate analytical expression ( 18 ) , @xmath178 . starting from @xmath179 k ,
the theoretical dependence has a weak extra peak ( marked by the arrow in fig .
1 ) at @xmath180 t , which comes from the longitudinal - mode contribution .
though the experimental magnetoresistance@xmath25 does not show a peak in this region , a precursor of such a feature is apparently visible as a flattening of the minimum between the peaks @xmath181 and @xmath182 at the temperatures above 4 k. as a function of inverse peak position @xmath183 extracted from the calculated magnetoresistance ( fig .
1 ) at 4 k. a linear fit for this dependence yields @xmath174 km / s .
the same kind of plot based on the experimental data is given in ref .
the positions of low - order peaks slightly deviate from the linear dependence because these peaks are formed by mixture of the oscillating contributions from different phonon modes.,width=302 ] it is important to compare the results of numerical calculation of the oscillating component of phonon - induced magnetoresistance with the results obtained from the analytical approach leading to eqs .
( 17)-(20 ) .
figure 3 ( a ) shows the partial contribution of high - energy phonon mode ( la ) calculated according to eq .
( 13 ) together with the corresponding contribution based on the expression @xmath184 , where @xmath144 and @xmath145 are given by eqs .
( 19 ) and ( 20 ) . in a similar way ,
3 ( b ) shows both numerical and analytical , based on eq .
( 18 ) , results for low - energy phonon modes ( ta ) .
the agreement between exact and approximate results becomes good in the low - field region corresponding to high - order magnetophonon resonances .
both numerical and approximate results in fig . 3 ( a ) demonstrate a beating pattern with a node at @xmath185 t. the analytical consideration reveals the origin of this beating , since it shows that the oscillations induced by la phonons are formed as a sum of two contributions , @xmath144 and @xmath145 , with slightly different frequencies .
k. ( a ) contribution of high - energy branch ( la ) .
thick ( black ) lines correspond to numerical calculation of @xmath109 using eq .
( 13 ) , and thin ( blue ) lines are the results obtained from the approximate analytical expressions ( 19 ) and ( 20 ) .
the inset shows beating pattern in the low - field region .
( b ) contribution of low - energy branches ( ta ) .
thick ( black ) and thin ( red ) lines correspond to numerical calculation using eq .
( 13 ) and to approximate analytical expression ( 18 ) , respectively .
the low - field region is shown in the inset.,width=347 ] figure 4 presents the phonon - induced magnetoresistance calculated for the gaas quantum wells with @xmath186 @xmath162 and @xmath187 nm studied in ref .
since the mobility of these samples ( @xmath188 @xmath152/v s at @xmath189 k according to the resistance at @xmath190 ) is much lower than that in ref .
17 , both experimental@xmath191 and theoretical magnetoresistance show a few low - order oscillations .
because of large density and , consequently , large fermi momentum in these samples , the main contribution to oscillating resistivity is caused by the deformation - potential interaction and comes from la phonons .
this is demonstrated by plotting partial contributions of different modes in fig .
4 . at the temperatures 15 - 25 k all acoustic - phonon modes
are fully activated , which means that the function @xmath192 is close to 1 , so the resistivity should linearly increase with temperature .
experimental plots of ref .
14 indeed show a linear increase for the background resistivity , while the amplitudes of the oscillation peaks depend on @xmath26 in a different way : the first peak is weakly modified by @xmath26 and the second peak is suppressed with increasing @xmath26 . such a behavior can be explained by a decrease of the quantum lifetime with increasing temperature owing to inelastic electron - electron scattering ( see ref .
17 and references therein ) . in the present formalism , this effect is described by adding the electron - electron scattering rate @xmath193 , where @xmath194 , to the inverse quantum lifetime@xmath195 @xmath31 entering the density of states in eq . ( 11 ) .
the result of the calculations improved in this way is also shown in fig .
4 . a reasonably good agreement with experimental data , as concerns the amplitudes of the oscillations and their temperature dependence , is achieved at @xmath196 .
k , the partial oscillating contributions from high - energy branch ( la ) and low - energy branches ( ta ) are also shown ( shifted for clarity ) .
the additional ( orange ) plots take into account temperature - induced decrease of the quantum lifetime and are in a better agreement with experimental data of ref .
14.,width=347 ] the positions of the calculated magnetoresistance peaks deviate from the experimental data , though this deviation is not strong . the fitting procedure similar to that shown in fig .
2 gives the characteristic velocity @xmath197 km / s , which is smaller than @xmath198 km / s determined in ref .
it is worth noting , however , that the positions of the peaks in the experiment ref .
14 are shifted to higher magnetic fields as the temperature increases .
though the reason of this shift is not clear , it signifies an increase of experimental velocity @xmath9 with temperature .
therefore , one may expect a better agreement of theory and experiment at lower temperatures .
this paper presents a microscopic theory of magnetophonon oscillations , also known as phonon - induced resistance oscillations ( piro ) , in 2d electron systems .
the phonon - induced resistivity @xmath80 is found by considering both deformation - potential and piezoelectric mechanisms of interaction of 2d electrons with bulk acoustic phonon modes . in the experimentally relevant situation of weak magnetic fields ( large filling factors ) , the expression for the resistivity ( or , eqivalently , for the phonon - assisted transport rate @xmath79 )
is determined by the electron - phonon coupling constants , phonon frequencies , and the density of electron states in magnetic field ; see eq .
. the essential feature of the calculations is that the anisotropy of the phonon spectrum is explicitly taken into account .
the theory is applied to quantum wells based on the materials of cubic symmetry and grown in the [ 001 ] crystallographic direction . for this case ,
an analytical expression for the oscillating part of the resistivity has been derived .
this expression is valid in the region of fields corresponding to high - order magnetophonon resonances , when the ratio of characteristic phonon frequency to cyclotron frequency is large .
the calculations are carried out for gaas quantum wells where piro have been observed experimentally .
this includes the cases of 2d electron gas with high density and moderate mobility@xmath191 and with lower density and very high mobility.@xmath25 the calculated magnetoresistance is in agreement with experimental data . as concerns the positions of the magnetoresistance peaks ,
the agreement is very good in the case of high - mobility systems.@xmath25 based on the results obtained , one may conclude that the bulk approximation for description of acoustic phonon modes and of their interaction with 2d electrons works reasonably good in application to magnetotransport in gaas quantum wells .
another important conclusion is that the velocity @xmath9 in the empirical equation ( 1 ) does not , in general , correspond to a certain phonon mode .
the reason for this is the complicated structure of the phonon spectrum .
there are three anisotropic phonon modes ( branches ) with different velocities and two mechanisms , deformation - potential and piezoelectric , of electron - phonon interaction .
relative contributions of these modes and of the interaction mechanisms depend on 2d electron density , temperature , and other parameters such as the width of the quantum well and growth direction .
this explains why the velocity @xmath9 determined by fitting the experimental piro peak positions is expected to vary in different experiments , especially when such a fit is based upon a few low - order peaks which have the highest amplitudes and , therefore , are best visible experimentally .
indeed , the low - order peaks are typically formed as a superposition of contributions from different modes , and even for each single mode there is a mixture of phonons with different velocities owing to the anisotropy .
the anisotropy influences not only the frequency but also the phase and the amplitude of the oscillations . on the other hand , in the region of high - order magnetophonon resonances
the resistance oscillations are described by a fixed set of phonon velocities determined here for the case of [ 001]-grown quantum wells ; see eqs .
( 17)-(20 ) .
the phases of these oscillations also become definite . moreover ,
if the density of 2d electrons is not large , so the piezoelectric - potential interaction is significant , the main contribution to magnetoresistance oscillations in [ 001]-grown wells is characterized by a single velocity @xmath199 corresponding to a ta mode propagating in the quantum well plane and polarized perpendicular to this plane .
this is the case of experiment ref .
17 , where many magnetoresistance oscillations have been observed owing to a very high electron mobility . for the wells whose growth axes are different from [ 001 ] the characteristic phonon velocities have to be different .
these cases are not studied here in detail because piro have not yet been observed in such wells .
nevertheless , based on the consideration presented in sec .
iii , the following general method for determination of characteristic phonon velocities in the wells of arbitrary growth axis is proposed .
take the surfaces of equal frequency @xmath200 defined for each mode in the @xmath201-space by the equation @xmath202 .
find the points ( numbered by the index @xmath203 ) where these surfaces touch the surface of the cylinder of radius @xmath15 and axis along the growth direction ( this cylindrical surface is given by the equation @xmath204 ) .
the set of frequencies @xmath205 ( or , equivalently , velocities @xmath206 ) corresponding to these points will give a set of oscillating harmonics with arguments @xmath207 in the magnetoresistance .
an equivalent procedure is the search for extrema of the function @xmath208 , as described in sec .
iii . to find which frequencies out of @xmath209 give the main contribution and to determine the phases and amplitudes of the oscillations , a detailed analysis is necessary . in the case of [ 001]-grown wells
a single ta mode gives the main contribution [ eq .
( 18 ) ] because the surface of equal frequency for this particular mode touches the cylinder not just in a finite number of points , but along the whole circumference of the cylinder at @xmath127 . to study the influence of different phonon modes on the magnetoresistance oscillations , it is desirable to carry out experiments in high - mobility 2d systems with densities varied in the range @xmath210 @xmath162
. if it is possible to reach high mobilities for the structures with growth axes different from [ 001 ] , measurements of the oscillating magnetoresistance in such systems would be also important for investigation of the phonon anisotropy effects discussed above .
the author hopes that the present theoretical work may stimulate such experiments .
the consideration given in this paper assumes the linear response regime , when both electron and phonon systems are close to equilibrium .
one may expect a more interesting and rich behavior of the phonon - induced resistance oscillations under nonlinear transport regime when the electric current through the sample increases .
studies in this direction are already undertaken.@xmath211 in addition to the work already done , more efforts , both experimental and theoretical , are necessary for better understanding of the mechanisms of phonon - induced oscillations and their interplay with the other microscopic mechanisms responsible for the magnetoresistance of 2d electrons at large filling factors .
a. a. bykov , d. p. islamov , a. v. goran , a. i. toropov , jetp lett .
* 87 * , 477 ( 2008 ) .
m. a. zudov , i.v .
ponomarev , a. l. efros , r. r. du , j. a. simmons , and j. l. reno , phys .
* 86 * , 3614 ( 2001 ) .
phonon - assisted scattering also leads to tempearature dependence of quantum lifetime .
however , the corresponding additional contribution to the inverse quantum lifetime is of the order of background phonon - assisted transport rate @xmath108 and is much smaller than @xmath31 caused by electron - impurity scattering .
|
a microscopic theory of the phonon - induced resistance oscillations in weak perpendicular magnetic fields is presented .
the calculations are based on the consideration of interaction of two - dimensional electrons with three - dimensional ( bulk ) acoustic phonons and take into account anisotropy of the phonon spectrum in cubic crystals .
the magnetoresistance is calculated for [ 001]-grown gaas quantum wells .
the results are in agreement with available experimental data .
apart from the numerical results , analytical expressions for the oscillating part of magnetoresistance are obtained .
these expressions are valid in the region of high - order magnetophonon resonances and describe the oscillating magnetoresistance determined by several groups of phonons polarized along certain high - symmetry directions .
|
the theorem of hohenberg and kohn @xcite concerning the existence of a universal energy density functional ( edf ) , and its subsequent implementation as a local density approximation ( lda ) , lead to a new qualitative approach to the study of electron systems , from atoms , to molecules , to condensed matter systems and macromolecules in particular and other fermionic systems in general .
even though neither hohenberg and kohn nor kohn and sham gave us recipes on how to construct the edf , various approximation schemes with increasing level of sophistication have been created .
moreover , the edf ideology has been extended to finite temperatures and finite excitation energies as well . however , essentially all of the implementations of the lda and edf have been limited so far to normal fermi systems , namely , systems with no pairing correlations .
there were two attempts to extend the lda to superconducting systems @xcite , however , the pairing field in this approach was still a nonlocal object .
one can present the argument that because electron superconductivity is phonon mediated , and since phonons have a spectrum limited by the debye frequency , such a genuinely nonlocal character of the electron pairing field is natural . in our opinion
this kind of argumentation is somewhat tenuous .
the normal part of the edf arises from coulomb interaction , which in itself has an infinite range .
nevertheless , a coherent lda approach to normal systems is possible . in recent works
@xcite we have been able to develop a genuinely local extention of the lda to systems with superfluid correlations and apply it to a number of nuclear and atomic systems .
besides the fact that a genuine local approach to pairing correlations within an lda la kohn and sham is certainly possible , such a framework is physically meaningful . in nuclear physics for example ,
the so called coherence length , or in other words the size of the cooper pair , is significantly larger than the radius of the nn - interaction .
the binding energy of the cooper pair , roughly equal to the pairing gap , is significantly smaller than the fermi energy .
thus a zero range approximation to the pairing interaction is definitely a meaningful thing to pursue , once one learns how to deal with the inherent short range / ultraviolet divergence characteristic to any local pairing field .
not the last among various arguments that one can bring forward , is the fact that our intuition is so much better in the case of local potentials , when compared to the case of nonlocal potentials .
this argument alone , together with the intrinsic simplicity of a local treatment will alone warrant the quest for a suitable , from a pure pragmatic point of view , local scheme , even if such a scheme would not be possible in principle .
we suggest a new acronym for the extention of the lda to superfluid systems , namely slda , standing for superfluid lda .
the starting point of the entire formalism is naturally the assumption that a local edf for superfluid systems exists , namely @xcite & & ( * r*)=_n[(*r*),(*r * ) ] + _ s[(*r*),(*r * ) ] , + & & ( * r * ) = _ i2|v_i(*r*)|^2 , + & & ( * r * ) = _ i 2| v_i(*r*)|^2 , + & & ( * r * ) = _ i v_i(*r*)^ * u_i(*r * ) , where @xmath0 $ ] is the normal contribution and @xmath1 $ ] is the superfluid counterpart . @xmath2 and @xmath3 are the normal density and kinetic energy densities and @xmath4 is the anomalous ( superfluid ) density , all expressed through the quasiparticle wave functions @xmath5 .
we have not shown explicitly the spin degrees of freedom ( so far we have limited ourselves to systems with @xmath6-wave pairing only ) .
the edf can and does depend on a number of other local densities , which for the sake of the simplicity of the presentation we choose not to display as well .
we assume that the normal part of the edf is known and we shall not discuss its origin and form . in most physical systems
the magnitude of the anomalous density is relatively small , which reflects the fact that in most cases the pairing is in the weak coupling regime .
one can then safely assume that , in nuclei for example , the superfluid edf is only quadratic in the anomalous density and in that case slda equations acquire the following structure ( shown here only for one kind of fermions ) : & & e_gs = d^3r \{_n[(*r*),(*r * ) ] + _ s[(*r*),(*r * ) ] } , + & & _
s[(*r*),(*r*)]:= -(*r*)_c(*r * ) = g__eff_(*r*)|_c(*r*)|^2 , + & & \ {
l [ h ( * r * ) - ] u_i ( * r * ) + ( * r * ) v_i ( * r * ) = e_i u_i ( * r * ) , + ^ * ( * r * ) u_i ( * r * ) - [ h ( * r * ) - ] v_i ( * r * ) = e_i v_i ( * r * ) , .
+ & & h ( * r * ) = - + u(*r * ) , + & & ( * r * ) : = -g__eff_(*r*)_c(*r * ) , + & & = - \ { 1 - } + & & _ c(*r * ) = _ e_i0^ e_c 2|v_i(*r*)|^2 , + & & _ c(_r _ ) = _
e_i0^ e_c v_i^*(*r*)u_i(*r * ) , + & & e_c + = + u(*r * ) , + & & = + u(*r * ) . ( nb in ref .
@xcite in the equation for the renormalized coupling constant there is a typo and the effective mass should be used as shown above . )
the slda equations for the quasiparticle wave functions have exactly the same structure as the hfb / bogoliubov - de gennes equations , for the trivial reason that the corresponding edf depends on one - body densities alone . unlike the hfb approximation , however , slda does in principle lead to the exact ground state energy and densities , up to gradient corrections .
the gradient corrections are additional terms in edf determined by @xmath7 and @xmath8 , which otherwise vanish identically in infinite homogeneous matter .
while the dependence of the edf on densities can be inferred from _ ab initio _ calculations of infinite homogeneous matter , the gradient corrections require additional input .
the uncertainties still existing in their determination are the source of the largest errors in various edf approaches and the main _ raison dtre _ for the letter `` a '' in the acronym lda .
as the vast experience accumulated over the years ( mainly in the study of a large number of electron systems ) amply shows , gradient corrections are never dominant and are indeed always ( relatively small ) corrections .
above @xmath9 is a cutoff energy , which should be chosen @xmath10 times the value of the fermi energy or larger .
in such a case any dependence of the results / observables on the value of the cutoff energy @xmath9 disappears . the only new element in the formalism , when compared to a formalism with an explicit energy cutoff , is the position and energy cutoff running coupling constant @xmath11 .
the main difference with the similar running coupling constants in quantum field theory ( qft ) for example , is the fact that @xmath12 depends on position as well , because we are dealing with inhomogeneous systems , unlike the qft case of particles interacting in vacuum , which is by default homogeneous .
the apparent quantum mechanical inconsistency that one can have at the same time a dependence of the running coupling constant on both position and energy ( more exactly on momentum ) cutoff is easily resolved if one remembers that as a matter of fact there is a clear separation of scales , similar to the separation of scales in the landau fermi liquids theory for example .
our knowledge of the normal part of the nuclear edf is more or less satisfactory @xcite . in our calculations of nuclear properties
we have used the so called sly4 interaction @xcite in order to generate @xmath13 $ ] and also fayans fandf@xmath14 @xcite , both of which where in somewhat different ways fitted to the canonical infinite matter results @xcite . as far as @xmath15 $ ] goes
, the theoretical knowledge is in a very unsatisfactory overall state .
infinite matter calculations made within the hfb / bcs framework lead to maximum pairing gaps of the order of 3 mev for @xmath16 and essentially vanishing pairing gaps at nuclear saturation densities . various `` correlation effects '' taken _ a posteriori _ into account are essentially never in agreement with each other , except for the fact that the maximum value of the gap is _ reduced _ to about 1 mev , a value which a number of people think is a reasonable one @xcite . in qualitative agreement with these results is a somewhat less known result in the nuclear community , established more than four decades ago . in
very dilute systems the bcs value for the pairing gap is incorrect and the actual value , which can be estimated quite accurately , is smaller by a factor of about 2.2 @xcite .
the situation becomes even more confusing in a way when one takes into account the fact that nuclei have a surface .
it was shown that surface modes lead to a significant _ enhancement _ of the pairing gap @xcite , thus just to the opposite effect .
nuclear phenomenology does not fare much better in this respect and one can find claims that the nuclear pairing energy has either a volume , or a surface or a surface + volume character @xcite .
thus the density dependence suggested by nuclear matter calculations ( weak pairing inside and somewhat stronger outside ) does not seem to be either confirmed or disproved in phenomenological studies of finite nuclei .
moreover , so far we really have no clue on whether pairing interaction is momentum dependent and/or energy dependent and/or isospin dependent .
just about the only thing we can state with some certainty is the fact that pairing correlations are indeed present in nuclei and on some kind of average the paring gap is about 1 mev or so . and since most of the phenomenological studies of the pairing effects are made with quite a number of restrictions , widely varying from one study to another , the more detailed information about pairing effects is often contradictory and thus can not be trusted . in our fully self - consistent treatment of nuclei , with an exact treatment of the continuum spectrum
as well , we have resorted to the most simple and meaningful approach , and we have decided to use a single bare coupling constant for both neutrons and protons . we have studied the case of @xmath17 pairing only so far . unlike many / most of the existing treatments available in the literature , our approach is fully consistent with the isospin symmetry of the nuclear forces . within this framework
we have been able to achieve an agreement with experiment for the single - nucleon and two - nucleon separation energies which compares extremely favorably with any of the previous calculations , see refs .
@xcite .
we have tried to determine whether the pairing coupling constant has any noticeable density dependence and were unable to find any .
we attribute this to the fact that a nuclear cooper pair is unable to resolve such fine details , since it effectively averages the pairing interaction over the entire nuclear volume . in a recent fully self - consistent analysis of
all known nuclear masses goriely _
et al._,@xcite have arrived independently at the same conclusion .
this particular aspect deserves a little more discussion .
often it is argued that the pairing strength among nucleons should be weaker / vanishing inside and reach its maximum at the nuclear surface and beyond .
such a behavior follows from a naive implementation of a `` local density approximation '' ( not to be confused with lda for edf ) to the pairing gaps as a function of density in infinite homogeneous nuclear matter @xcite .
a better alternative to the naive local density approximation is either the thomas - fermi approximation .
a discussion of the merits and demerits of the thomas - fermi approximation , which indeed would provide only an approximate solution to the slda equations , was performed by grasso and urban @xcite in the hfb limit . in ref .
@xcite one can find as well a comparison between the renormalization scheme proposed by @xcite us and its precursor @xcite .
in the weak coupling limit , the value of the hfb / bcs pairing gap is quite accurately given by the emery s formula @xcite .
when one introduces the low - density corrective pre - exponential factor @xcite , the corrected hfb / bcs emery s formula for the pairing gap becomes ( if @xmath18 ) & = & ( ) ^7/3 ( - ) + & & ( ) ^7/3 ( ) . in the naive local density approximation , when the value of the pairing gap at a given point inside a nucleus is given by the value of the pairing gap in infinite homogeneous matter at the corresponding local density , a strong density dependence can arise if the @xmath19 effective pairing interaction in momentum representation @xmath20 has a strong momentum dependence .
a strong momentum dependence arises if the range of the effective nn pairing interaction is relative large , namely of the order of @xmath21 or larger . in the usual hfb approximation
a density dependence of this type does also lead to a large nonlocality of the pairing field .
notice , however , that in slda the pairing field is local and has no momentum dependence . in slda the momentum dependence of the pairing field can arise only if one were to introduce a dependence of the edf on a new anomalous density , similar to the normal kinetic energy density @xmath22 , namely _
( * r * ) = _ i v_i(*r*)^ * u_i(*r * ) .
this simple argument , based on naive local density approximation , that a noticeable density dependence should be present in nuclei , is not without its merits and it is indeed surprising that our and goriely _ et al._,@xcite analyses could not find any convincing trace of such a behavior . even if one is willing to accept our argument that _ a spatially large and weakly bound cooper pair can not really resolve such small details _ and it effectively averages the strength of the pairing coupling constant over the entire nuclear volume , it is not clear why such an average is independent of the atomic number .
clearly the ratio surface / volume is not constant over the periodic table .
it is worth noticing as well , that so far , in the somewhat limited study we have performed , we never found a reason to introduce an isospin dependent coupling , even though we have definitely covered the nuclear region where such a dependence was advocated rather strongly by others .
moreover , to stress it again , both proton and neutron pairing correlations have been described with the same coupling constant , as indeed one would expect ( up to relatively small coulomb and csb forces corrections ) .
even though we have not been able to identify any isospin dependence of the pairing edf , such a dependence could exist and it was suggested in various ways before .
very often however , various authors have violated the isospin symmetry of the pairing edf , either invoking phenomenological arguments , or simply in order to be able to obtain a better description of various nuclear properties , in particular masses @xcite .
we suspect that one can reconcile to some extent ( not fully though ) the lack of isospin symmetry in refs .
@xcite for example , in a rather simple manner , by simply noticing that so far experimentalists have been able to create more neutron rich nuclei than proton rich nuclei .
it can be shown that = 0.1473 , where the average is computed over all measured nuclear masses with @xmath23 used in ref .
one can then easily show that the following superfluid edf & & _
s(_n,_p,_n,_p ) + & & = g(_n,_p ) + f(_n,_p ) , + & & g(_n,_p)= g(_p,_n ) < 0,f(_n,_p)= f(_p,_n ) > 0 , + & & - 0.39 will reproduce the fact than on average the proton pairing is stronger than the neutron pairing , in agreement with the otherwise isospin violating treatment of pairing correlations in refs .
the above nuclear superfluid edf has two parts .
the first part , proportional to @xmath24 , is indeed isospin symmetric .
however , the second part , which is proportional to @xmath25 , is only charge symmetric .
it is highly debatable , however , whether one can really accept a charge symmetric only ( as opposed to an isospin symmetric one ) contribution to the nuclear superfluid edf , merely for the sake of improving the agreement of the calculated masses for example with the experimental values .
so far we could not find a suitable candidate for a nuclear superfluid edf , which could in principle directly couple the proton and neutron superfluids and which is not more than quadratic in the anomalous densities
. there have been suggestions for superfluid nuclear edf in literature @xcite , quartic in character . recently
this question has been raised again @xcite , in order to determine whether protons in neutron stars form a type i or type ii superconductor , following the observations of a long period precession in isolated pulsars @xcite , which apparently do not support the standard picture of protons being a type ii superconductor .
we find the superfluid edf suggested in ref .
@xcite , however , very hard to reconcile with our knowledge of the pairing correlations at the corresponding densities @xcite .
a relatively simple in structure and very easy to implement lda to edf of fermion systems with superfluid correlations has been developed , which in itself apparently represents the first genuinely local extension of the kohn - sham ideology to such systems .
this superfluid lda ( slda ) has been applied by us so far to study single - nucleon and two - nucleon separation energies of a relatively large number of nuclei ( more than 200 ) with a surprisingly high accuracy .
we have used the simplest possible ansatz for the superfluid energy density , compatible with all basic nuclear symmetries , and have used standard forms for the normal energy density part of the nuclear edf .
namely , we have used a superfluid edf characterized by a single , density independent and universal bare coupling constant . and even though we did not try to obtain the best overall description of these basic energy nuclear properties , which are affected most significantly by the pairing correlations , our results proved to be of a better quality than any of the previous results we are aware of .
of course , there is no guarantee that by extending the analysis to more nuclei the accuracy of this simple approach will survive unscathed .
but there is plenty of room for `` improvements , '' if such would be needed .
there is no question in our minds that a more sophisticated form of the superfluid edf is going to be needed in order to improve even further the quality of the agreement between theory and experiment .
in particular , it is very likely that an explicit isospin dependence of the pairing couplings will eventually emerge .
there are a number of questions our results left so far unanswered , for various reasons . in particular , it is still unclear whether a contribution to the nuclear superfluid edf exists , which couples directly the proton and neutron superfluids .
the microscopic calculations available in literature , concerning the dependence of the pairing gaps on density leave so far a lot to desire , as no consensus seem to have emerged to what is the true value of the pairing gap for example in neutron matter @xcite .
moreover , one can expect that the values and the density dependence of the pairing gaps could be entirely different in symmetric nuclear matter , if one were to naively extrapolate the arguments of heiselberg _
et al._,@xcite to finite densities , and we should expect an enhancement of the pairing correlations when compared to simple bcs / hfb calculations , thus an effect opposite to that established so far in pure neutron matter . the dependence of the pairing gap on the isospin composition of the matter is still unknown microscopically .
things are however somewhat worse , as an additional density dependence should appear in finite nuclei . since nuclei have clear and well defined surface collective modes and as it appears that their contribution to the pairing gaps are about 50% or so @xcite , this leads us to be believe that gradient density corrections to the superfluid edf could be rather large .
some of these aspects of the nuclear superfluid edf become particularly acute in neutron stars , as at higher than nuclear saturation densities the pairing in @xmath26- and @xmath27-waves becomes important .
for example , it is a matter of current debate whether at such densities protons are a superconductor of type i or of type ii @xcite , and the answer to this question can change a lot of the neutron stars physics .
the slda formalism developed by us has been applied to other systems as well , the vortex state in low density supefluid neutron matter @xcite , the vortex in a superfluid dilute atomic fermi gas @xcite , overall properties of superfluid correlations in such systems @xcite , 2-dimensional quantum dots @xcite , in many cases leading to new qualitative findings .
the range of phenomena to which slda has been applied so far , with pairing gaps spanning almost twenty orders of magnitude , is a measure of its flexibility and relevance .
it is worth mentioning that there are at least two cases in which the slda approach can be implemented already in a fully controlled manner .
namely , in the low density regime both the normal edf and the pairing gap are known to a high degree of accuracy @xcite .
unlike the hf approach , which would be valid in the leading order in this case , the hfb / bcs fails in this limit , however , slda works .
the hf approximation is accurate for electrons in high density regime as well . as a matter of fact
we are not aware of any physical system in any regime where the unadulterated hfb / bcs approximation will have a satisfactory accuracy .
another extremely interesting and universal regime is that of a dilute system , but with an infinite scattering length @xcite , a regime which is nowadays routinely achieved in dilute atomic gases .
quite accurate calculations of such homogeneous systems became available recently @xcite and subsequently the slda approach was implemented for inhomogeneous systems , specifically to describe the vortex state , see ref .
@xcite .
it is fair to conclude that our knowledge of the pairing properties of both finite nuclei and infinite neutron and nuclear matter are in a very unsatisfactory state .
the heavily phenomenological trend , which dominated the study of pairing nuclear properties over the last forty years left us with no answers to most of the questions discussed here .
it is our hope that the existence of a theoretically consistent slda framework should be of significant help in settling some of these aspects .
a fundamental aspect of the nuclear pairing problem concerns the calculations of nuclear masses .
the emphasis in nuclear physics so far was to generate , following one strategy or another , see refs .
@xcite , the best possible nuclear mass formula .
our attitude should change , from trying to describe these masses with the best possible accuracy , to trying to understand why , when using a meaningful formalism , which incorporates our best established physical input , we still fail at some level or another .
we should realize that by now we have at our disposal close to 2,500 measured masses and we should treat this as an object in its entirety , from which we may learn new physics perhaps , by using a framework consistent with what we have learned so far about nuclear interactions .
there is a long list of people from whom the first author ( ab ) has benefitted in various ( sometimes indirect ) ways over the last few years , when work on this project took place , namely : p.f .
bedaque , g.f .
bertsch , r.a .
broglia , j. carlson , j. dobaczewski , b. friman , e.k.u .
gross , n. hayashi , p .- h .
heenen , e.m .
henley , yu .
a. litvinov , w. nazarewicz , yu.n .
novikov , p. magierski , j. meng , g. mnzenberg , v.r .
pandharipande , j.m .
pearson , p. pizocherro , p .-
reinhard , c. scheidenberger , b. spivak , s. stringari , a. svidzinsky , n. takigawa , e. vigezzi , h.a .
weidenmller and some of the referees of our papers .
this work would have been almost impossible without the financial support of doe under the contract de - fg03 - 97er41014 .
p. hohenberg and w. kohn , phys . rev . *
136 * , b864 ( 1964 ) ; w. kohn and l.j .
sham , phys .
* 140 * , a1133 ( 1965 ) ; r.m .
dreizler and e.k.u .
gross , _ density functional theory : an approach to the quantum many body problem _ , ( springer , berlin , 1990 ) ; r.g .
parr and w. yang , _ density functional theory of atoms and molecules _ , clarendon press , oxford ( 1989 ) ; w. kohn , rev .
phys . * 71 * , 1253 ( 1999 ) .
p. ring , prog .
. phys . * 37 * , 193 ( 1996 ) ; m. bender _
et al . _ , phys . lett . * b 515 * , 42 ( 2002 ) ; p.g .
et al . _ , nucl - th/0012095 ; d.j .
dean and m. hjorth jensen , rev .
* 75 * , 607 ( 2003 ) ; m. bender _
et al . _ , rev .
phys . * 75 * , 121 ( 2003 ) ; p. ring and p. schuck , _ the nuclear many body problem _ , springer , new york ( 1980 ) .
j. dobaczewski _
c * 53 * , 2809 ( 1996 ) ; j. dobaczewski _ et al .
_ , nucl.phys .
* a 693 * , 361 ( 2001 ) ; s. sakakihara and y. tanaka , nucl . phys . * a 691 * , 649 ( 2001 ) ; m. bender _
j. a * 8 * , 59 ( 2000 ) ; t. niksic _ et al .
_ , phys.rev .
c*66 * , 024306 ( 2002 ) .
d. zavischa _ et al .
_ , , phys .
* b 185 * , 299 ( 1987 ) ; m. apostol _
et al._,europhys .
lett . * 4 * , 197 ( 1987 ) and nucl
. phys . * a 470 * , 64 ( 1987 ) o. dumitrescu and m. horoi , nuovo cimento * a 103 * , 635 ( 1990 ) ; m. horoi , phys .
c * 50 * , 2834 ( 1994 ) .
|
we present a concise account of our development of the first genuine local density approximation ( lda ) to the energy density functional ( edf ) for fermionic systems with superfluid correlations , with a particular emphasis to nuclear systems .
|
regge calculus ( rc ) provides a natural framework for the description of discrete spacetimes @xcite .
it has been applied to problems ranging from quantum gravity to classical relativity @xcite . in recent years
rc has been used in quantum gravity to construct emergent spacetime models @xcite and to investigate the spin - foam approach and its perturbative regime @xcite . in particular
, @xcite utilizes a pre - geometric framework wherein the spacetime is an emergent structure from an underlying quantum system .
the unitary operators of the discrete quantum system act as source terms for the ensuing spacetime . if an approach of this kind is to utilize rc as the coupling to the underlying quantum dynamics , it is incumbent upon us to understand the nature of the coupling of matter and fields to a rc lattice spacetime .
most early attempts to include matter and fields in rc concentrated on spacetimes with a high degree of symmetry @xcite or in specific cases @xcite .
a more general approach was given by @xcite ; however , this approach introduced a simplex - based action which does not appear to be natural to us given the vertex - based description indicated by the contracted bianchi identity . a general theory of the coupling of source terms to rc has not been fully developed .
it is the goal of this manuscript to provide the first steps to construct a simplicial representation for the stress - energy of the non - gravitational sources .
to do so we first examine the extent to which a simplicial lattice can guarantee the conservation of energy - momentum .
we then develop more fully the contracted bianchi identity on the lattice .
this identity , when augmented by the einstein equations , yields the conservation of energy and momentum .
we know no shorter route to derive the contracted bianchi identity than by using the topological tautology that the boundary of a boundary is zero , i.e. the boundary - of - a - boundary principle ( bbp ) .
the bbp appears twice over in each of nature s four fundamental interactions , once in its 1 - 2 - 3-dimensional form , and once in its 2 - 3 - 4-dimensional form @xcite .
the bbp has been used in rc to obtain a discrete version of the contracted bianchi identity @xcite .
however , the interpretation of this conservation law has been a source of some debate particularly over the exactness of the identity . while the topological principle itself is exact and thoroughly studied in rc @xcite , the transition from a continuum to a discrete spacetime forces one to apply these topological identity to non - infinitesimal rotations .
unlike the infinitesimal rotation operators in the continuum , finite rotations do not ordinarily commute . the transition from the continuum to the discrete case must be handled with care .
we emphasize that the derivation presented here will not ordinarily produce an exact identity due to the non - commuting nature of finite rotations .
nevertheless what one loses in exactness one gains in simplicity .
in particular , the integrated einstein tensor is doubly projected along its edge and this allows one to write down the contracted bianchi identity as a kirchhoff - like conservation principle . these identity
are second - order convergent @xcite and valid locally at any event in a spacetime .
the contracted bianchi identity for rc has clear implications for the coupling of energy - momentum to the lattice as well as to our understanding of diffeomorphism invariance in rc .
furthermore , if we expect the quantization of spacetime to produce an inherently discrete spacetime , then grasping the meaning of the bbp in a discrete theory becomes essential to understanding the quantum theory of gravity .
rc serves naturally as an underlying framework since simplicial spacetimes provide one of the most elegant and universal descriptions of discrete spacetime @xcite . in sec .
[ bbp ] we review the bbp and its role in the fundamental forces of nature .
the importance of this identity stems from its purely topological foundation .
the cartan construction of the moment of rotation trivector in rc is reviewed before applying the bbp directly to the simplicial lattice in sec .
[ bbprc ] .
we conclude in sec . [ discussion ] with our future plans to couple a generic stress - energy tensor to the geometric content of the regge lattice .
in any fundamental field theory ( electrodynamics , yang mills , general relativity ) the conservation of source is introduced in such a way that it is satisfied for any field .
this is equivalent to saying that it does not impose any restrictions on the field itself but rather puts constraints on the source of the field ( charge in electrodynamics , energy
momentum in general relativity ) .
this feature is conditioned only by the requirement that the field is described as the curvature of a connection on the appropriate vector bundle that is responsible for the correct implementation of the field symmetries @xcite .
the universality of this feature follows from the fact that it is induced by ( but not totally reduced to ) the simple topological identity that the boundary - of - a - boundary is equal to zero @xcite .
application of this principle to spacetime is achieved by associating to it a chain complex ( say by simplicial or any other triangulation ) with the standard boundary operator based on the rules of orientations of the boundaries . as an example
, we can examine a discrete representation spacetime wherein the spacetime geometry is tiled by 4-dimensional polytopes .
the geometry interior to each of these infinitesimal polytopes is irrelevant and , for pictorial representations , can be thought of as flat minkowski geometry .
let us examine one of these polytopes , @xmath0 , which is the local neighborhood of an event , @xmath1 .
this polytope is bounded by 3-dimensional polyhedra ( figure [ fig:4v ] ) .
any two adjacent polyhedra on the boundary of @xmath0 share a common 2-dimensional face .
in other words , in this 4-dimensional region of spacetime no 2-dimensional polygonal faces are exposed . in general relativity ,
any flow of stress - energy ( or equivalently the dual of the cartan moment of rotation ) into one of the 3-dimensional bounding polyhedra is exactly compensated by an equal flow of stress - energy ( cartan moment of rotation ) out of an adjacent polyhedron .
this guarantees conservation of source in @xmath0 .
, of the circumcentric dual ( voronoi ) spacetime .
it is dual to a vertex , @xmath1 , of the simplicial ( delaunay ) spacetime .
this 4-polytope is bounded by six polygons ( shown exploded off into the perimeter of the polytope ) .
these 4-polytopes are ordinarily not simplexes nor are their bounding polyhedra .
the orientation of @xmath0 induces an orientation on each of its polyhedral faces , @xmath2 .
the orientation of each polyhedron consequently induces an orientation on each its polygonal faces .
however , each 2-face is shared by two polyhedra thereby inducing equal and opposite orientations on it . in this sense none of these polygonal faces
are exposed and their orientations cancel .
this is the origin of the bbp principle in its 2 - 3 - 4 dimensional form .
, height=288 ] for each @xmath0 , one would like to sum over each of its unexposed 2-dimensional boundaries the meeting place of two of the polyhedral boundaries of @xmath0 ( figure [ fig : bbp ] ) .
these polyhedral boundaries induce opposing orientations on each of the 2-dimensional faces .
therefore when one sums over all of the 2-boundaries of all the 3-boundaries , two contributions are found for each polygon each of equal magnitude but opposite orientations .
these identically cancel one another leaving the boundary - of - a - boundary identically equal to zero .
is parallel transported around the area adjoining the two 3-volumes , then the vector will ordinarily come back rotated . when the area is associated with the left 3-volume ,
the vector @xmath3 comes back rotated as @xmath4 , but when the area is associated with the right 3-volume it will come back as @xmath5 .
the rotation in both cases is in the same plane and rotated by the same amount but in opposite directions of rotation .
, width=288 ] in applications to continuum field theories the boundary of a boundary relation of the chain complex is translated into the relation co - boundary of a co - boundary of the dual de rham co - chain complex of differential forms .
the exterior derivative acts as the co - boundary operator @xcite .
this duality is established by adding rotations caused by parallel translations of vectors around the 2-faces ( or cartan moments of these rotations ) of an infinitesimal 3-simplex ( or 4-simplex for moments of rotations ) .
these rotations are expressed as products of riemannian curvature tensors on each face and the oriented element of area associated with the face .
this operation , when applied to finite structures , is ambiguous and can not be performed in a consistent way .
the ambiguity arises for two reasons : finite rotations do not commute , and tensor quantities are being computed at different points ( on different faces ) and then added .
these difficulties disappear in the infinitesimal limit .
in particular the application of 1 - 2 - 3 dimensional bbp in general relativity reduces to computing the sum of rotations caused by parallel transport around all 2-faces of a 3-simplex .
when expanded in taylor series with respect to displacements along the edges @xcite it produces terms of second and the third order ( the higher orders are of no interest in this computation ) .
the ambiguity caused by parallel transporting tensor quantities to a common point ( necessary for addition ) introduces errors of the fourth and higher order .
these errors can be neglected .
the requirement that the second order term vanishes leads to the conclusion that the riemannian curvatures on the faces of the 3-simplex are not linearly independent , while the requirement that the third order term vanishes implies the ordinary bianchi identity .
likewise , application of 2 - 3 - 4 dimensional bbp amounts to adding cartan moments of rotation over all 3-faces of a 4-simplex ( or polytope ) .
the taylor series expansion proceeds as before , with terms up to the third order disappearing because of relations imposed by the 1 - 2 - 3 dimensional form of the bbp .
the errors generated by the ambiguity of parallel translation are now of the fifth and higher order .
the contracted bianchi identity arises from the fourth order term of the expansion .
regge calculus is based on a lattice of flat 4-dimensional simplexes that form a curved piece - wise flat manifold .
the curvature is concentrated as conical singularities on each of the co - dimension 2 triangular hinges . to define the curvature we need to associate an `` area - of - circumnavigation '' to each triangular hinge .
this provides a finite area over which we can distribute the curvature .
fortunately , the circumcentric dual lattice has been shown to arise naturally in rc and provides an appropriate area @xcite .
correspondingly , it has been shown that the circumcentric 3-volumes naturally define the moment - of - rotation operators and discrete rc equations @xcite .
finally , we postulate here that the circumcentric 4-volume ( @xmath0 ) dual to a vertex ( @xmath1 ) defines a natural domain to apply the cartan bbp in its 2 - 3 - 4-dimensional form .
consequently the bbp in rc becomes the `` co - boundary of the co - boundary '' principle , although the geometric underpinnings are exactly the same . in this section
we derive the discrete form of the contracted bianchi identity .
we begin by emphasizing the central role that the cartan moment - of - rotation trivector and the circumcentric dual lattice play in this derivation . in particular , we begin by re - expressing the familiar regge equations in the cartan prescription .
this leads naturally to a kirchhoff - like identity at each vertex inherently linked with the topological boundary - of - a - boundary identity .
we conclude the section with an analysis of the convergence properties of the identity with the typical lattice edge length @xmath6 . to begin the derivation we follow e. cartan and examine the moment of rotation trivector .
the dual to this trivector generates the regge equation as well as the rc version of the einstein tensor .
recall that there is one regge equation for each edge , @xmath6 , in the simplicial spacetime geometry , @xmath7 where the sum is taken over each triangular hinge @xmath8 sharing edge @xmath6 , @xmath9 is the interior angle of @xmath8 opposite @xmath6 , and @xmath10 is the deficit angle of hinge @xmath8 .
the cartan moment - of - rotation trivector is defined through a moment arm ( @xmath11 ) reaching from a fulcrum to a rotation bivector .
each triangular hinge @xmath8 in the simplicial spacetime has an associated rotation bivector ( @xmath12 ) located at the circumcenter @xmath13 of the hinge @xmath8 .
the orientation of @xmath12 is in a 2-plane , @xmath14 , perpendicular to the hinge , @xmath8 .
the bivector is formed by two unit vectors separated by the usual rc deficit angle , @xmath10 .
it is convenient to locate the fulcrum at one of the two endpoints edge @xmath6 .
we denote this fulcrum vertex as @xmath1 , and by construction it is one of the vertices of hinge @xmath8 .
this freedom of choice is guaranteed by the ordinary bianchi identity as we show below .
this is in contrast to previous derivations of the regge equations using the cartan approach where the fulcrum was taken halfway along edge @xmath6 @xcite . with the fulcrum at @xmath1
we can decompose the moment arm associated with hinge @xmath8 into two vectors ( fig .
[ fig : hinge ] ) , @xmath15 where @xmath16 is the vector from the fulcrum @xmath17 to point @xmath18 , located at the center of edge @xmath6 .
this is also the center of three - dimensional circumcentric polyhedron @xmath19 , defined to be dual to edge @xmath6 .
the other component of the moment arm , @xmath20 , is the vector from @xmath18 to the circumcenter @xmath21 of the hinge .
this gives us two vectorial contributions to the moment arm ; one ( @xmath22 ) is common to all 2-dimensional faces @xmath14 of the dual polyhedron , @xmath2 , and another ( @xmath20 ) is distinct for each of these 2-dimensional faces .
the contribution common to all faces of @xmath23 can be factored out of the sum of moments of rotations , so that @xmath24 the resulting sum over all rotations around @xmath19 is simply the ordinary bianchi identity for rc @xcite , @xmath25 in eqs .
[ fulcrummove]-[obi ] the sum over the hinges , @xmath26 , sharing edge @xmath6 could have equally been taken over the bounding polygons , @xmath14 , of the dual polygon , @xmath27 . there is a one - to - one correspondence between the @xmath8 and @xmath14 . using this approximate bianchi identity
we are justified in removing the common contribution to the moment arm in our sum over the moments - of - rotation .
we see that the ordinary bianchi identity allows us to freely choose the position of the fulcrum .
a natural choice for the fulcrum is the vertex @xmath1 and we use @xmath28 to denote the dual 4-polytope to this vertex .
then each edge , @xmath6 , emanating from vertex , @xmath29 , has the moment arm , @xmath16 , which is directed along edge , @xmath6 .
since each edge @xmath6 is dual to a corresponding 3-polytope , @xmath2 , the effective moment arm is @xmath30 which is the segment from the center of the edge ( @xmath31 ) to the circumcenter ( @xmath13 ) of the hinge ( figure [ fig : hinge ] ) .
the circumcentric 3-volume @xmath2 dual to edge @xmath6 has 2-dimensional boundaries dual to each of the hinges @xmath8 ( right )
. the parallel transport of a vector around the perimeter of these dual areas will result in a net rotation by an angle equal to the deficit angle , @xmath10 , associated with the hinge , @xmath8 .
the moment of rotation is given by a moment arm @xmath32 wedge the rotation associated with the parallel transport around the dual area .
however , the first term does not contribute as it is equal to zero by the ordinary bianchi identity . on a given hinge , the effective moment arm is the vector from the edge to the center of rotation , i.e. the circumcenter of the hinge @xmath13 , which has length @xmath33 .
, width=384 ] we are now in a position to explicitly reconstruct the regge equation associated to an edge @xmath6 , and to construct the corresponding regge - einstein tensor . to define the moment of rotation trivector associated to hinge @xmath8 and edge @xmath6 we needed both the moment arm and the rotation bivector .
parallel transport of a unit vector around the 2-dimensional face @xmath14 dual to the hinge @xmath8 returns a unit vector rotated by an amount equal to the deficit angle ( @xmath10 ) associated with the hinge .
furthermore , the rotation bivector lies in the plane @xmath14 , perpendicular to the hinge
. the dual of the einstein tensor is expressible in terms of the moment of rotation trivector @xcite , @xmath34 where the hodge dual only acts in the space of values , i.e. on the moment of rotation trivector . in rc
the moment of rotation trivector consists entirely of the parallelepiped formed by the moment arm @xmath35 and the two vectors defining the rotation bivector , @xmath36 which lies in the plane orthogonal to the triangular hinge @xmath8 .
this hinge is defined by the vectors @xmath37 and @xmath38 , and has area @xmath39 .
the star dual of the parallelepiped returns a vector of length @xmath10 and parallel to edge @xmath37 .
we can now construct the moment - of - rotation trivector .
the dual moment of rotation associated with a hinge @xmath8 containing the edge @xmath6 is @xmath40 the _ total _ dual moment of rotation over the voronoi 3-volume @xmath2 is then found by adding contributions from all hinges which share the edge @xmath6 , @xmath41 in the cartan description of einstein s theory @xcite the einstein tensor associated with a three dimensional region is the dual of the total moment of rotation tri - vector per unit three - volume .
the two components of the einstein tensor describe the orientation of the three volume , and the orientation of the moment of the rotation tri - vector . in regge calculus
there is one equation per edge @xmath6 , as can be seen when the cartan moment of rotation is calculated over the voronoi three - volume @xmath2 @xcite .
the orthogonality between the simplicial ( delaunay ) lattice and its circumcentric dual ( voronoi lattice ) yields an einstein tensor which is doubly projected along edge @xmath6 .
that is , @xmath42 which is directed along edge @xmath6 and has magnitue @xmath43 . combining equations ( [ momrot ] ) , ( [ totalmoment ] ) and ( [ reggemoment ] ) establishes the relationship between the regge equations and the _ integrated _ simplicial einstein equations @xcite , @xmath44 this effectively defines the simplicial einstein tensor @xmath45 at edge @xmath6 .
finally , we note that the simplicial einstein tensor along the edge @xmath6 , constructed using the sum of moments of rotations for the dual 3-volume @xmath2 , is simply the geometric portion of the familiar regge equation , @xmath46 as shown above , the regge equation and corresponding simplicial einstein ( or regge - einstein ) tensor is naturally defined relative to the dual polyhedron @xmath2 .
it is therefore natural to define the 4-dimensional polytope @xmath47 dual to the vertex @xmath17 as the domain upon which we apply the cartan bbp in its 2 - 3 - 4 dimensional form . to demand no net creation of source ( @xmath48 ) in this spacetime region
is to embody the essence of the contracted bianchi identity .
as in the continuum ( sec . [ bbp ] )
, we can provide a finite sum ( `` integral '' ) representation of the contracted bianchi identity associated to the dual polytope @xmath49 by summing over its polyhedral 3-boundaries @xmath23 . by construction , each polyhedral 3-boundary of the dual polytope @xmath50 is dual to one of the edges @xmath6 of the simplicial lattice emanating from vertex @xmath1 .
this defines the domain of integration for the bbp .
this completes two steps toward the bbp in rc by defining both the domain and the integrand ( sec .
[ cartan - rc ] ) for the bbp in rc .
the final step in deriving an expression for the conversation of moment of rotation in rc is achieved by summing over the dual 3-volumes @xmath23 that bound the 4-volume @xmath0 , which is dual to vertex @xmath1 . however , care must be taken in evaluating this sum . despite our choice of a common fulcrum
, we have still decomposed the total moment arm into two vectors ( one strictly in the tangent space of the vertex @xmath1 , and one at the center of edge @xmath6 ) .
yet , rc provides a simple solution to what could be problematic . both of these vectors lie in the tangent space of their associated hinge @xmath8 . as such , the decomposition of the total moment arm becomes the standard decomposition of a vector in flat minkowski spacetime .
summation over terms at a common point can be achieved in two equivalent , but separate approaches ; ( 1 ) by parallel transporting the effective moment arm prior to inclusion in the moment of rotation trivector , or ( 2 ) parallel transporting the net moment of rotation trivector .
we will consider the second approach . in rc the integrated einstein tensor ( [ erequations ] )
is not only evaluated along the edge @xmath6 , it is also directed along @xmath37 . since
each simplicial edge @xmath6 is by definition a geodesic in the lattice , any vector parallel transported along @xmath37 will maintain a constant angle with respect to @xmath37 .
we take advantage of this property and individually transport each of the rc moment - of - rotation trivectors from the center of their respective edges to the vertex @xmath1 which is common to all of these edges .
we are then free to sum these moment - of - rotation vectors at @xmath1 . repeating this procedure across the lattice
yields a 4-vector identity at each and every vertex @xmath1 , @xmath51 this is the simplicial form of the net moment of rotation at vertex @xmath1 and must vanish by the 2 - 3 - 4 dimensional form of the bbp .
however , as we have mentioned , the finite rotation operators do not ordinarily commute .
this is important because we must apply our rotations in a given order .
nevertheless , the non - commutativity of the rotation operators can be made as small as one wishes by suitably refining the lengths of the simplicial lattice . here
, suitable refinement of the lattice is taken in the sense described in @xcite where constant curvature barycentric subdivision is employed to refine the edge lengths by introducing new simplicial blocks and distributing curvature over the new subdivision of the simplexes .
under such refinements , the commutators for rotations scale as the deficit angles squared . moreover ,
the deficit angles scale as the edge length squared as can be seen via their relation to the curvature @xmath52 .
consequently , the deficit angles scale as @xmath53 and the commutators for rotations scale as @xmath54 .
this second order convergence is the origin of the approximation being implemented .
therefore , @xmath55 this is the rc formulation of the contracted bianchi identity .
the first term in this expression scales with @xmath56 , since the deficit angles @xmath10 scale as @xmath57 .
the contracted bianchi identity is not identically zero because small , finite , rotations do not necessarily commute .
consequently , the final term scales with both the edge length @xmath6 and the rotation commutator @xmath58 $ ] , yielding an overall @xmath59 behaviour in the error term . two features are apparent in the discrete contracted bianchi identity . first , it has the form of a kirchhoff - like conservation law .
the analysis presented in this manuscript completes the derivation of this kirchhoff - like property of the contracted bianchi identity .
we have reduced the results of previous calculations from a non - local , boundary - valued sum to a vertex - based conservation equation .
this was accomplished by utilizing the freedom we have in choosing the fulcrum for each of the moment of rotation trivectors , here we have chosen the vertex @xmath1 common to all of the faces of the dual polytope @xmath60 .
equivalently , this can be understood by our ability to parallel transport the moment of rotation trivector from the midpoint @xmath61 the edge @xmath62 to the vertex @xmath1 .
no higher order corrections than corrections already discussed in this manuscript are introduced .
it is this understanding of the relation between the contracted bianchi identity gives rise to conservation of source which is vital to understanding how source can be coupled to the lattice .
second , the appearance of a 4-vector identity at each vertex signals that there are exactly four `` approximate '' diffeomorphic degrees of freedom per vertex in the simplicial lattice .
this last point has been important for understanding the dynamical degrees of freedom in rc @xcite , and the resulting approximate diffeomorphism freedom has been utilized to solve the initial value problem @xcite . and their dual 3-volumes , @xmath2 .
the first step towards the kirchhoff - like conservation principle is constructing the total moment of rotation for each of the dual 3-volumes , @xmath2 . on the right is a depiction of the flow of moment - of - rotation , or equivalently flow of stress - energy along each of the edges , @xmath6 meeting at the vertex @xmath1 .
the flow of moment - of - rotation entering or leaving vertex , @xmath1 , is conserved to second order in the lattice spacing , @xmath6 . since the einstein equations , and their rc equivalent , equates the moment - of - rotation with the stress - energy , this contracted bianchi identity can be viewed as a circuit - like conservation law . here
the `` wires '' of the circuit are the edges of the simplicial lattice , and the `` current '' in each of the wires is the doubly - projected stress - energy tensor @xmath63 along the given edge @xmath37 emanating from vertex @xmath1 .
, title="fig:",height=192 ] and their dual 3-volumes , @xmath2 .
the first step towards the kirchhoff - like conservation principle is constructing the total moment of rotation for each of the dual 3-volumes , @xmath2 . on the right is a depiction of the flow of moment - of - rotation , or equivalently flow of stress - energy along each of the edges , @xmath6 meeting at the vertex @xmath1 .
the flow of moment - of - rotation entering or leaving vertex , @xmath1 , is conserved to second order in the lattice spacing , @xmath6 . since the einstein equations , and their rc equivalent , equates the moment - of - rotation with the stress - energy , this contracted bianchi identity can be viewed as a circuit - like conservation law . here
the `` wires '' of the circuit are the edges of the simplicial lattice , and the `` current '' in each of the wires is the doubly - projected stress - energy tensor @xmath63 along the given edge @xmath37 emanating from vertex @xmath1 .
, title="fig:",height=192 ]
rc has been a predominately vacuum theory of gravitation with only specialized applications to spacetimes with non - gravitational sources @xcite .
our approach can provide a consistent and generic coupling of matter to the simplicial lattice .
the cornerstone of our approach relies on the nature of conservation principles of the lattice .
we know of no better way to understand the conservation than through the bbp .
this application of the bbp has led to an approximate conservation of moment - of - rotation for rc .
the conservation of moment - of - rotation takes a form which is ideally suited for applications to matter : the contracted bianchi identity in rc becomes a kirchhoff - like conservation principle along the edges emanating from a specific vertex . with this and einstein s equations ,
we obtain an approximate conservation equation for energy - momentum on the lattice , @xmath64 in particular , the doubly projected stress - energy along the edges emanating from a vertex must sum to zero , to at least second order in the length scale of the lattice . while not exact , this gives the interpretation of a kirchhoff - like conservation principle for the geometry , and ( with einstein s equations ) the flow of energy and momentum ( figure [ fig : kirchhoff ] ) . as a result
, we obtain a set of vertex - based constraints for edge - based expressions that constrain energy - momentum .
this exercise indicates that energy - momentum is naturally wired to the simplicial lattice at each vertex and is naturally wired to each edge in its coupling with the simplicial field equations . for applications of rc to pre - geometric quantum spacetime
, one must necessarily formulate an appropriate stress - energy tensor arising from the quantum dynamics . for applications to classical spacetimes a simplicial form of the stress - energy tensor
must be constructed from the non - gravitational sources .
this work indicates that the stress - energy will most naturally be expressed as a vertex - based tensor , and that its coupling to the rc equations will be through its double projection on the edges of the lattice .
we will explore this coupling in future work .
we were fortunate to have been exposed to the deep insights and motivations of john archibald wheeler .
his valuable contributions to this work made it possible to provide such a clear geometric model of conservation of energy - momentum in a discrete spacetime .
|
simplicial lattices provide an elegant framework for discrete spacetimes . the inherent orthogonality between a simplicial lattice and its circumcentric dual yields an austere representation of spacetime which provides a conceptually simple form of einstein s geometric theory of gravitation . a sufficient understanding of simplicial spacetimes has been demonstrated in the literature for spacetimes devoid of all non - gravitational sources .
however , this understanding has not been adequately extended to non - vacuum spacetime models .
consequently , a deep understanding of the diffeomorphic structure of the discrete theory is lacking .
conservation laws and symmetry properties are attractive starting points for coupling matter with the lattice
. we present a simplicial form of the contracted bianchi identity which is based on the e. cartan moment of rotation operator .
this identity manifest itself in the conceptually - simple form of a kirchhoff - like conservation law .
this conservation law enables one to extend regge calculus to non - vacuum spacetimes and provides a deeper understanding of the simplicial diffeomorphism group .
|
in the analysis of the black hole evaporation , one usually assumes that a horizon is formed in a collapse process , and examines the evaporation and entropy in the static black hole @xcite-@xcite . in this paper
we try to build a self - consistent model which describes both formation and evaporation of a black hole including the back reaction from the hawking radiation .
that is , we solve the semi - classical einstein equation in a self - consistent manner : @xmath2 where @xmath3 contains the contribution from both the collapsing matter and the hawking radiation . from the solution we can investigate whether a horizon and singularity are formed or not .
we first consider a null shell as the collapsing matter and construct the geometry by connecting the inside flat metric and the outside outgoing vaidya metric on the shell .
note that particle creation generally occurs in a time - dependent gravitational potential , and especially , the hawking radiation can appear without a horizon @xcite .
we invent a formula that evaluates the energy flux of such a process .
then we obtain self - consistent equations which determine time evolution of the shell and the radiation .
the solution shows that the radiation stops , the horizon and singularity appear , and the black hole remains forever .
next we analyze the case where a continuous null matter collapses and discuss the mechanism of the hawking radiation .
it has an onion - like internal structure and evaporates gradually from the outermost part .
then we write down a self - consistent stationary solution in the heat bath .
it has neither a macroscopically large horizon nor singularity .
we first explain the general idea for construction of a geometry which describes a black hole from formation to evaporation .
next we propose a simple model .
suppose that a gravitational collapse forms a schwarzschild black hole as in the left of fig.[fig : general ] .
if we take time reversal , the existing black hole goes back to the flat spacetime as in the center of fig.[fig : general ] .
then , if we cover the inside of the horizon and the singularity by pasting a collapsing matter , we obtain a geometry which describes both the formation and evaporation as in the right of fig.[fig : general ] .
note that whether this picture is realized or not depends on the dynamics . therefore we need to make some model and solve it concretely .
we will consider the following model .
when we take a null shell as the collapsing matter , the inside spacetime is flat : @xmath4 as a simple model of the outside metric , we take the outgoing vaidya metric @xcite : @xmath5 where @xmath6 is the bondi mass and the only non - zero component of the einstein tensor is @xmath7 where the null energy condition implies @xmath8 .
this is the general spherically symmetric metric which satisfies @xmath9 and @xmath10 except for @xmath11 .
, where @xmath12 is the schwarzschild radius of the null shell , we can take @xmath13 because most partial waves with @xmath14 of the radiation do not go through their own centrifugal barrier in @xmath15 .
next the incoming flux can be neglected there because of the boundary condition that any energy flow does not come from infinity except for the shell .
furthermore if we consider only massless fields , we can assume @xmath9 because the weyl anomaly vanishes approximately in @xmath16 . at @xmath17 ,
the ingoing flow and @xmath18 can exist with @xmath14 , but we assume to neglect them for the simplest model . therefore we can consider the conditions . in this sense the outgoing vaidya metric represents the outgoing radiation without the gray - body factor . ]
note that the coordinate @xmath19 must be the same in the both side , because it is defined as the radius of 2-sphere and there is no room to rewrite @xmath20 . on the other hand ,
@xmath21 is related to @xmath22 as @xmath23 where @xmath24 is the locus of the null shell .
this comes from the fact that @xmath25 is an ingoing null geodesic in the both sides .
thus a simple model is given by connecting the outgoing vaidya metric and the flat metric with the null shell as in fig [ fig : vaidya ] .
we call it one - shell model
. approaches its own schwarzschild radius @xmath26 exponentially . ] here we analyze the locus of the null shell @xmath27 for a given function @xmath26 .
@xmath27 is determined by the condition : @xmath28 this equation tells that the shell will approach its own schwarzschild radius in the time scale @xmath29 if @xmath26 changes so slowly that the time scale in which @xmath26 changes significantly , @xmath30 , is much larger than @xmath12 , that is , @xmath31 then , in the region @xmath32 , we can replace @xmath25 in the denominator with @xmath12 and solve it as @xmath33 where @xmath34 is a positive constant . here
the term @xmath35 means that as the shell approaches to its schwarzschild radius in the time scale of @xmath36 , the radius reduces by the evaporation .
( see fig [ fig : vaidya ] . )
therefore the shell can not catch up with the radius completely as long as @xmath8 , but it approaches to @xmath37 finally we investigate the surface energy - momentum tensor @xmath38 on the shell . using the barrabes - israel null - shell formalism @xcite , we estimate @xmath39 where @xmath40 is the four vector of an observer ( @xmath41 ) , @xmath42 is the ingoing radial null vector which is taken as @xmath43 in the vaidya metric and @xmath44 in the flat space , and @xmath45 is the metric on the 2-sphere ( @xmath46 ) . the fact that @xmath47 implies that the work done by the shell as it contracts is transformed to the hawking radiation .
thus this model is consistent in energetics .
time evolution of this model depends on the functions @xmath26 and @xmath27 , so we will investigate their dynamics in the following sections .
we will here construct a flux formula @xmath48 which , at @xmath49 , estimates energy flow from the black hole : @xmath50 in the heisenberg picture , we use the eikonal approximation , the point - splitting regularization and only the s - wave to obtain @xmath51\equiv \frac{\hbar}{8\pi}\left\{u , u\right\ } , \label{j}\ ] ] whose form is the same as the schwarzian derivative .
the derivation is given in the [ sec : der_flux ] .
note that we can also derive the planck distribution without horizon ( see [ sec : planck ] ) .
first we test the formula in the case without back reaction , that is , in the geometry obtained by connecting the schwarzschild metric and the flat space . in this case , from , @xmath27 becomes @xmath52 where @xmath12 becomes constant completely .
then the flux is estimated as @xmath53 where @xmath54 .
this is the same as thermal radiation from a one - dimensional black body with the temperature @xmath55 . in this sense ,
the flux formula is consistent with the usual result @xcite .
note that this result is the same as the stefan - boltzmann law except for the coefficient . from , , and ,
we have obtained the self - consistent equations which determine the dynamics of the one - layer model , that is , @xmath26 and @xmath27 : @xmath56 @xmath57 , \label{sceq2}\ ] ] where @xmath58 is the planck length .
now we consider the collapse of a null shell by using the one - shell model and investigate whether it evaporates or not . the numerical result of and
is shown in fig .
[ fig : shell ] . in the case of the null shell .
the initial conditions are given by with @xmath59 and @xmath60 . ] here we have chosen the initial conditions given by @xmath61 where @xmath62 is a number , and we assume @xmath63 .
the shell does not evaporate completely , and a horizon and singularity appear .
this asymptotic behavior does not depend on the detail of the initial conditions .
we can understand why the radiation stops in the following manner .
let s recall the estimation of the hawking radiation on the geometry with @xmath64 ( see ) . in that case
, only the term proportional to @xmath65 contributes to the formula .
however , now the term @xmath66 appears in @xmath25 , , and the exponential factor will damp for large @xmath21 .
therefore , @xmath25 becomes @xmath67 asymptotically as in .
because @xmath68 is at most of order @xmath69 , we can approximate @xmath70 then and can be solved as @xmath71 where @xmath72 and @xmath73 are integration constants , and @xmath72 is small and positive . from @xmath74 corresponds to @xmath75 , so shows that @xmath26 will not necessarily vanish as @xmath76 .
thus we have seen that a collapsing null shell with radius @xmath25 radiates for a while , but it stops and the radius @xmath25 almost stays at the schwarzschild radius @xmath12 .
then the horizon and singularity appear .
a single shell does not evaporate completely even if the back reaction from the hawking radiation is taken into account .
we discuss the case where a continuous null matter collapses ( see fig . [
fig : cont ] ) . .
the redshift factor is required to compare physics between different shells . ]
let s consider a shell .
the metric just outside the shell is given by @xmath77 where @xmath78 is the schwarzschild radius corresponding to the total energy of the lower shells , and @xmath79 is the time coordinate for the shell .
the locus of the shell @xmath80 follows @xmath81 and behaves as in the one - shell model : @xmath82 furthermore , for the shell , the flux formula holds , and we have @xmath83 where @xmath84 represents the energy flux measured at infinity if the shells outside did not exist .
here we have introduced @xmath85 degrees of freedom . in the case of the standard model , @xmath86 in the energy region higher than the weak scale .
note that introducing @xmath85 corresponds to replacing @xmath87 with @xmath88 .
we will show that the hawking radiation is emitted from each shell , but only shells near the outermost one are relevant because of the large redshift .
first we introduce a notation @xmath89 which represents distance between the shell @xmath90 and the schwarzschild radius @xmath91 . here
we assume that the last term in has already damped for each shell , so that we have @xmath92 then the energy flux for each shell depends only on its schwarzschild radius and does not have an explicit @xmath79-dependence : @xmath93 from , @xmath94 also becomes a function of @xmath91 : @xmath95 now we consider the junction condition of the adjacent shells . by looking at each shell from the both side ( see fig . [
fig : cont ] ) , we obtain @xmath96 if @xmath97 is small , we get @xmath98 by integrating it , we obtain the redshift factor between @xmath91 and @xmath99 for finite distance : @xmath100 from this equation , the quantity @xmath101 can be rewritten as @xmath102 where and have been used .
thus we have found that for each shell @xmath103 which is independent of a concrete form of @xmath104 . on the other hand , by expressing @xmath105 in terms of @xmath106 , we can express the energy flow in as @xmath107 from and , we obtain @xmath108 by substituting and iteratively , we have @xmath109 where @xmath110 , which would be the temperature measured at infinity if the shells outside did not exist .
thus , any shell can emit the hawking radiation if the shells are continuously distributed so that we can use .
this result does not depend on the behaviour of the shells outside the one we are considering . from , and , @xmath111 is determined as @xmath112 by considering the outermost shell , we find that the total hawking radiation is given by @xmath113 which coincides with the result for the static schwarzschild geometry . then by applying to the outermost shell , we obtain the time evolution of the size of the black hole : @xmath114 we can also show that the energy spectrum of the radiation follows the planck distribution ( see [ sec : planck ] ) .
therefore this black hole evaporates completely as is usually expected .
however , our model describes how it happens more precisely .
actually the black hole evaporates gradually from the outermost shell as if one peels off an onion . here
we will check that the total radiation is equal to the sum of the radiation from each shell .
first let s estimate the radiation from the region between @xmath91 and @xmath115 as depicted in fig .
[ fig : cont ] .
if there were no shells outside this region , the radiation is estimated as @xmath116\left ( 1-\frac{da}{\rho(a ' ) } \right)\\ & = & \frac{d f ( a ' ) } { da'}da + f(a')\frac{da}{\rho(a')}\\ & \approx & \frac{da}{2a'},\end{aligned}\ ] ] where @xmath117 is neglected as a higher term , and is used . by using this and the redshift factor , the sum of radiation from each layer is estimated as @xmath118 here the dominant contribution in the integration comes from the outermost thin region with a width about @xmath119 ( see ) . although each shell radiates , the outermost region gives the main contribution because of the large redshift .
we consider the case that the black hole is put in the heat bath with the hawking temperature of the outermost shell for long time so that holds for each shell .
it is not difficult to calculate the metric for this stationary geometry , and we obtain ( see [ sec : metric ] ) @xmath120 this expression is valid for @xmath121 and smoothly connected to the schwarzschild metric at @xmath122 .
this metric does not have a horizon . here
@xmath123 is the time of the flat space at infinity , which is related to the time around the origin @xmath124 as @xmath125 this means that @xmath124 is so much redshifted that @xmath124 is almost frozen from viewpoint of an observer at infinity .
note that this geometry has been obtained self - consistently , so the classical limit @xmath126 does not exist .
this metric does not have a large curvature compared with @xmath127 in the region @xmath128 if @xmath85 is sufficiently large , @xmath129 : @xmath130 the singularity around the origin @xmath131 is controllable in the sense that it can be removed by introducing a small shell surrounding the origin . for example , suppose a small shell with @xmath132 comes first , and next , it grows to a large size with @xmath133 in the heat bath .
then the outside region @xmath134 is described as the stationary metric , while the center shell is the schwarzschild black hole with the radius @xmath135 which does not evaporate forever as in the case of the one - shell model .
therefore we have a horizon and singularity around the origin , but their size is small . here
we make a comment on the weyl anomaly .
the trace of the einstein tensor is given by @xmath136 because classically the energy - momentum tensor of null shells should be traceless , this should be identified with the weyl anomaly .
actually , if we use the formula of the weyl anomaly for n scalar fields @xcite , we obtain @xmath137 which agrees with up to numerical coefficients .
therefore the self - consistent solution obtained by the eikonal approximation already contains the effect of the weyl anomaly .
we have solved the semi - classical einstein equation in a self - consistent manner .
we have built a model which describes a black hole from formation to evaporation including the back reaction from the hawking radiation .
we consider null matter collapse and assume that the geometry is obtained by connecting the matter region and the outgoing vaidya metric . using the eikonal approximation , we have found a formula that gives the energy flux of the particle creation in a dynamical geometry .
then we have obtained the self - consistent equations which determine time evolution of the collapsing matter and radiation . as the first example
, we have analyzed the case where a single shell collapses and solved it numerically and analytically .
the shell does not evaporate completely , and a horizon and singularity appear .
this is not a thermodynamic object but a stable one in the sense that it can not be formed nor evaporated adiabatically in a heat bath .
next we have discussed the case where a continuous null matter collapses .
then the hawking radiation occurs not only from the surface but also from the inside .
however , because of the large redshift , the radiation is emitted substantially only from the region around the surface .
this black hole evaporates as is usually thought .
we then have put it in a heat bath and found the stationary metric .
it dose not have a macroscopically large horizon or singularity . by introducing a small shell around the origin
, this singularity can be controlled .
the metric automatically takes into account the effect of the weyl anomaly .
there remain some open problems .
our stationary solution has neither horizon nor singularity , so the information inside the hole must come back after evaporation . however , we do nt understand the mechanism clearly yet . for example
, suppose that we throw a newspaper into the stationary black hole described by our metric .
it will behave like another null shell going to the hole as it approaches the surface .
clearly its energy will be transformed into the hawking radiation by our mechanism .
however , the radiation itself comes from the quantum field on the past infinity , or the vacuum .
how will the information of the newspaper come back ?
a clue to this problem is that we have taken the expectation value of the energy - momentum tensor @xmath138 in our self - consistent equations , which might correspond to the coarse - graining procedure in the ordinary statistical mechanics . on the other hand ,
if we put our black hole in a heat bath with the temperature equal to the hawking temperature of the outermost shell , it is completely stationary . in this sense
, our black hole can be regarded as a thermodynamic object having this temperature and its entropy is given by the area law .
we do nt claim that the information problem is solved , but our black hole does not have a macroscopically large horizon and singularity . the small singularity around the origin would be resolved by string theory . if it is the case , the system is completely well - defined .
the work of y.m . is supported by the jsps research fellowship for young scientists .
the work of y.y . is supported by the jsps research fellowship for young scientists and by the grant - in - aid for the global coe program `` the next generation of physics , spun from universality and emergence '' from the ministry of education , culture , sports , science and technology ( mext ) of japan .
we will here derive the flux formula by taking only the s - wave and using the eikonal approximation . from and , we estimate @xmath139 at @xmath49 in the one - shell model ( see fig .
[ fig : vaidya ] ) .
first we investigate the behavior of a massless scalar field at @xmath49 in the schwarzschild metric : @xmath140 where @xmath141 .
the action of the field @xmath142 on this metric is @xmath143 where we have decomposed the field into partial waves @xmath144 and introduced the new coordinate @xmath145 and the effective potential for each partial wave with angular momentum @xmath146 as @xmath147 this implies that only the s - wave survives at @xmath49 because partial waves with @xmath148 have to tunnel their own centrifugal barrier with the rate @xmath149 . then let s consider the wave equation for scalar field @xmath142 on the vaidya metric : @xmath150 where @xmath151 is the laplacian for angular directions . here
we take only the s - wave @xmath152 and use the eikonal approximation @xmath126 .
then we get @xmath153 therefore , for the outgoing modes , we obtain the equation : @xmath154 next in this approximation we consider time evolution of the heisenberg operator @xmath155 at @xmath49 in the collapsing spacetime .
( see fig .
[ fig : flux ] . ) before and after the collapse in the eikonal approximation . ] before the collapse , the field is given by the spherical waves : @xmath156 which corresponds to the field on the flat space . here
the vacuum is defined as the minkowski vacuum : @xmath157 after the collapse , the field becomes , from , @xmath158 where @xmath159 is any increasing function of @xmath21 .
we are here using the eikonal approximation , so the phase remains constant on the outgoing mode : @xmath160 where @xmath22 is the time coordinate in the flat space inside the shell , and at the second equality we have used the junction condition on the locus of the shell .
let s estimate the flux based on the above analysis .
we use the point - splitting regularization technique to subtract the divergence @xcite : @xmath161 , \label{t_uu1}\ ] ] where the time @xmath21 is after the collapse , and the r - dependence is not explicitly written because @xmath49 .
first we introduce @xmath162 and expand the equation with respect to @xmath163 . by using , the first term in
is estimated as @xmath164}\\ & = & \hbar \int^{\infty}_0 \frac{d\omega}{2\pi}\omega [ \dot f+\ddot f\epsilon+\dddot f \epsilon^2/2+\cdots ] [ \dot f-\ddot f\epsilon+\dddot f \epsilon^2/2+\cdots]e^{-i\omega[2\dot f\epsilon+\dddot f\epsilon^3/3+\cdots ] } \\ & = & \hbar \int^{\infty}_0
\frac{d\omega}{2\pi}\omega [ \dot f^2-\epsilon^2(\ddot f^2-\dot f\dddot f)+{\cal o}(\epsilon^4 ) ] e^{-2i\omega \dot f\epsilon}[1-i\omega \dddot f\epsilon^3/3+{\cal o}(\epsilon^5 ) ] \\ & = & -\frac{\hbar } { 8\pi \epsilon^2}+\frac{\hbar } { 8\pi}\left [ \frac{\ddot f^2}{\dot f^2}-\frac{2\dddot f}{3\dot f}\right]+ { \cal o}(\epsilon^2),\end{aligned}\ ] ] where @xmath165 . in the same way , the second term in is estimated as @xmath166 thus , by using , we obtain the flux formula for @xmath167 as @xmath51\equiv \frac{\hbar}{8\pi}\left\{u , u\right\},\ ] ] which is .
we emphasize that the planck distribution can be obtained even if the geometry has no horizon .
all that is necessary is that the affine parameters on the null generators of past and future null infinity are related exponentially @xcite . in this appendix
, we will show that in our model , the expectation value of the number of the particle creation takes the form of the planck distribution with the hawking temperature @xmath168 , in which @xmath26 changes so slowly that holds .
we start with reviewing the standard calculation of the hawking radiation .
we consider the state in the heisenberg picture that is annihilated by the positive frequency operators in the past infinity @xmath169 : @xmath170 because the profile of the wave is modified by the gravitational potential , the positive frequency operators in the future infinity @xmath171 is a superposition of the positive and negative frequency operators in the past infinity @xmath172 , @xmath173 then in the future infinity the number operator takes the non - trivial value @xmath174 the coefficient @xmath175 is given by the klein - gordon inner product : @xmath176 where @xmath177 is the wave function of the outgoing mode on the future null infinity and @xmath178 is that on the past null infinity . as in [ sec : der_flux ] , we will use the eikonal approximation for the s - wave . then , from and , @xmath179 then becomes @xmath180 here we need the relation between @xmath21 and @xmath22 .
\(1 ) in the case of a single shell , we can use @xmath181 and @xmath182 where @xmath183 is a time when the exponential factor remains . here
note that @xmath34 is positive because @xmath184 .
then we obtain @xmath185 thus can be evaluated as @xmath186 where the irrelevant phase factor is dropped . here
the contribution from @xmath21 away from @xmath187 is negligible @xcite because the only interval @xmath188 $ ] contributes to the integral , where @xmath189 is a constant @xmath190 . after performing the @xmath21-integration
, we obtain @xmath191 where the irrelevant phase factor is omitted .
\(2 ) in the case of the asymptotic region of the continuous matter , we first expand @xmath26 around @xmath183 which is a time in the region : @xmath192 where for simplicity we have normalized the hawking radiation as @xmath193 .
then the redshift factor is estimated as @xmath194 thus we obtain @xmath195 where @xmath196 , and have @xmath197 this is different from in the sign of the exponentials , but this integral leads to almost the same result : @xmath198 for the both cases , using the formula @xmath199 we obtain @xmath200 where @xmath201 by considering a wave packet around @xmath187 , we arrive at the planck - distributed hawking radiation with temperature @xmath202 : @xmath203
we will derive the stationary metric .
the metric represents a vicinity around a point @xmath204 just outside a shell .
@xmath90 is so close to @xmath91 that , from , @xmath205 from , and , @xmath206 is also estimated as @xmath207 by using these , the time coordinate @xmath79 is related to that around the origin @xmath22 as @xmath208 now we connect it to the outside metric , that is , the schwarzschild metric at the outermost shell @xmath122 : @xmath212 comparing this with at @xmath213 , we obtain the relation @xmath214 therefore we reach the metric @xmath215 j. d. bekenstein , phys . rev .
2333 ( 1973 ) . s. w. hawking , commun .
phys . * 43 * , 199 ( 1975 ) [ erratum - ibid . *
46 * , 206 ( 1976 ) ] .
w. g. unruh , phys .
d * 14 * , 870 ( 1976 ) .
s. w. hawking , phys .
d * 14 * ( 1976 ) 2460 .
j. b. hartle and s. w. hawking , phys .
d * 13 * , 2188 ( 1976 ) .
g. w. gibbons and s. w. hawking , phys .
d * 15 * , 2752 ( 1977 ) .
d. n. page , hep - th/9305040 .
s. d. mathur , lect .
notes phys .
* 769 * , 3 ( 2009 ) [ arxiv:0803.2030 [ hep - th ] ] .
a. strominger and c. vafa , phys .
b * 379 * , 99 ( 1996 ) [ hep - th/9601029 ] .
j. m. maldacena , jhep * 0304 * , 021 ( 2003 ) [ hep - th/0106112 ]
. s. d. mathur , fortsch .
phys . * 53 * , 793 ( 2005 ) [ hep - th/0502050 ] .
s. w. hawking , phys .
d * 72 * , 084013 ( 2005 ) [ hep - th/0507171 ] .
a. ashtekar and m. bojowald , class .
* 22 * , 3349 ( 2005 ) [ gr - qc/0504029 ] .
e. bianchi , arxiv:1204.5122 [ gr - qc ] .
a. almheiri , d. marolf , j. polchinski and j. sully , arxiv:1207.3123 [ hep - th ] .
l. susskind , arxiv:1208.3445 [ hep - th ] .
e. bianchi , arxiv:1211.0522 [ gr - qc ] .
t. vachaspati , d. stojkovic and l. m. krauss , phys .
d * 76 * , 024005 ( 2007 ) [ gr - qc/0609024 ] . c. r. stephens , g. t hooft and b. f. whiting , class .
* 11 * , 621 ( 1994 ) [ gr - qc/9310006 ] . c. barcelo , s. liberati , s. sonego and m. visser , jhep * 1102 * , 003 ( 2011 ) [ arxiv:1011.5911 [ gr - qc ] ] ; phys . rev .
d * 83 * , 041501 ( 2011 ) [ arxiv:1011.5593 [ gr - qc ] ] .
p. c. vaidya , proc .
indian acad .
sci . a * 33 * , 264 ( 1951 ) .
w. a. hiscock , phys .
d * 23 * , 2813 ( 1981 ) ; phys . rev .
d * 23 * , 2823 ( 1981 ) .
f. girotto and a. saa , phys .
d * 70 * , 084014 ( 2004 ) [ gr - qc/0406067 ] . c. barrabes and w. israel , phys .
d * 43 * , 1129 ( 1991 ) .
e. poisson , _ a relativist s toolkit : the mathematics of black hole mechanics _ , ( cambridge university press , cambridge uk , 2004 ) .
|
we construct a self - consistent model which describes a black hole from formation to evaporation including the back reaction from the hawking radiation . in the case where a null shell collapses , at the beginning the evaporation occurs , but it stops eventually , and a horizon and singularity appear . on the other hand , in the generic collapse process of a continuously distributed null matter , the black hole evaporates completely without forming a macroscopically large horizon nor singularity . we also find a stationary solution in the heat bath , which can be regarded as a normal thermodynamic object .
[ cols= " > " , ] * a self - consistent model of + the black hole evaporation * hikaru kawai@xmath0 , yoshinori matsuo@xmath1 and yuki yokokura@xmath0 _ @xmath0 department of physics , kyoto university , kitashirakawa , kyoto 606 - 8502 , japan + @xmath1 kek theory center , high energy accelerator research organization(kek ) , + oho 1 - 1 , tsukuba , ibaraki 305 - 0801 , japan _
|
let @xmath0 be a sequence of independent random variables such that for all @xmath1 , @xmath2 and for some @xmath3 and @xmath4 for integers @xmath5 , @xmath6 .
the classic bernstein inequality ( cf .
@xcite , page 855 ) says that , in this situation , for all @xmath7 and @xmath8@xmath9 moreover ( cf .
@xcite , theorem b.2 ) , its maximal form also holds ; that is , we have @xmath10 it turns out that , under a variety of assumptions , a sequence of not necessarily independent random variables @xmath0 will satisfy a generalized bernstein - type inequality of the following form : for suitable constants @xmath11 , @xmath12 , @xmath13 and @xmath14 for all @xmath15 , @xmath7 and @xmath16 , @xmath17 where , for any choice of @xmath18 , we denote the partial sum @xmath19 here are some examples . [ ex1 ]
let @xmath0 be a stationary sequence satisfying @xmath20 and .
for each integer @xmath7 set @xmath21 and @xmath22 . assume that for some @xmath23 we have @xmath24 for all @xmath7 .
statuleviius and jakimaviius @xcite and saulis and statuleviius @xcite prove that the partial sums satisfy ( [ assumpgamma ] ) with constants depending on a bernstein condition on the moments of @xmath25 and the particular mixing condition that the sequence may fulfill .
in fact , all values of @xmath26 are attainable .
their bernstein - type inequalities are derived via a result of @xcite relating cumulant behavior to tail behavior , which says that for an arbitrary random variable @xmath27 with expectation @xmath28 , whenever there exist @xmath29 , @xmath30 and @xmath31 such that its cumulants @xmath32 satisfy @xmath33 for @xmath34 then for all @xmath35 @xmath36 in example [ ex1 ] , @xmath37 and @xmath38 for some @xmath39 .
doukhan and neumann @xcite have shown , using the result in ( [ br ] ) , that if a sequence of mean zero random variables @xmath40 satisfies a general covariance condition , then the partial sums satisfy ( [ assumpgamma ] ) .
refer to their theorem 1 and remark 2 , and also see @xcite .
assume that @xmath0 is a strong mixing sequence with mixing coefficients @xmath41 , @xmath7 , satisfying for some @xmath42 , @xmath43 .
also assume that @xmath2 and for some @xmath44 for all @xmath1 , @xmath45 .
theorem 2 of merlevde , peligrad and rio @xcite implies that for some constant @xmath46 for all @xmath16 and @xmath7 , @xmath47 with @xmath48 and where @xmath49 to see how the last example satisfies ( [ assumpgamma ] ) , notice that for any @xmath50 there exists a @xmath51 such that for all @xmath16 and @xmath7 , @xmath52 thus ( [ assumpgamma ] ) holds with @xmath53 for suitable @xmath11 , @xmath12 and @xmath13 . for any choice of @xmath54 define@xmath55 we shall show , somewhat unexpectedly , that if a sequence of random variables @xmath56 satisfies a bernstein - type inequality of the form ( assumpgamma ) , then , without any additional assumptions , a modified version of it also holds for @xmath57 for all @xmath15 and @xmath7 .
[ t1 ] assume that , for constants @xmath11 , @xmath12 , @xmath13 and @xmath58 , inequality ( [ assumpgamma ] ) holds for all @xmath59 and @xmath60 . then for every @xmath61 there exists a @xmath46 depending only on @xmath62 , @xmath63 and @xmath64 such that for all @xmath7 , @xmath15 and @xmath16 , @xmath65 [ ra ] notice that though @xmath66 , @xmath67 can be chosen arbitrarily close to @xmath68 . [ r1 ] theorem [ t1 ] was motivated by theorem 2.2 of mricz , serfling and stout @xcite , who showed that whenever for a suitable positive function @xmath69 of @xmath70 , positive function @xmath71 defined on @xmath72 and constant @xmath73 , for all @xmath74 and @xmath75 , @xmath76then there exist constants @xmath77 and @xmath46 such that for all @xmath78 and @xmath75 , @xmath79earlier , mricz @xcite proved that in the special case when @xmath80 one can choose arbitrarily close to @xmath81 by making @xmath46 large enough .
this inequality is clearly not applicable to obtain a maximal form of the generalized bernstein inequality .
[ r ] we do not know whether there exist examples for which ( assumpgamma ) holds for some @xmath82 and @xmath83 .
however , since the proof of our theorem remains valid in this case , we shall keep it in the statement .
[ r2 ] the version of theorem [ t1 ] obtained by replacing everywhere @xmath84 remains true with little change in the proof .
[ r3 ] theorem [ t1 ] also remains valid for sums of banach space valued random variables with absolute value @xmath85 replaced by norm @xmath86 .
[ statistics ] in statistics , maximal exponential inequalities are crucial tools to determine the exact rate of almost sure pointwise and uniform consistency of kernel estimators of the density function and the regression function .
the literature in this area is huge .
see , for instance , @xcite and the references therein .
these results only treat the case of i.i.d .
dependent versions of our maximal bernstein inequalities should be useful to determine exact rates of almost sure consistency of kernel estimators based on data that possess a certain dependence structure .
in fact , some work in this direction has already been accomplished in section 4.2 of @xcite . to carry out
such an application in the present paper is well beyond its scope .
theorem [ t1 ] leads to the following bounded law of the iterated logarithm .
[ c1 ] under the assumptions of theorem [ t1 ] , with probability @xmath81 , @xmath87 [ r5 ] in general , one can not replace `` @xmath88 '' by `` @xmath89 '' in ( [ blil ] ) .
to see this , let @xmath90 , @xmath91 be a sequence of independent random variables such that @xmath90 takes on the value @xmath28 or @xmath81 with probability @xmath92 and @xmath91 are independent standard normals
. now define @xmath93 , @xmath94 it is easily checked that assumption ( [ assumpgamma ] ) is satisfied with @xmath95 @xmath96 , @xmath97 and @xmath98 when @xmath99 the usual law of the iterated logarithm gives with probability @xmath81 , @xmath100 whereas , when @xmath101 the above limsup is obviously @xmath102 this agrees with corollary [ c1 ] , which says that with probability @xmath81 the limsup is @xmath103 . however , we see that with probability @xmath92 it equals @xmath104 and with probability @xmath92 it equals @xmath28 .
theorem [ t1 ] is proved in section [ sec2 ] and the proof of corollary [ c1 ] is given in section [ sec3 ] .
the case @xmath97 is a special case of theorem 1 of @xcite .
therefore we shall always assume that @xmath83 .
choose any @xmath105 we prove our theorem by induction on @xmath106 .
notice that by the assumption , for any integer @xmath107 we may choose @xmath108 to make the statement true for all @xmath109 .
this remark will be important , because at some steps of the proof we assume that @xmath106 is large enough .
also , since the constants @xmath110 , @xmath68 , @xmath63 and @xmath111 in ( [ assumpgamma ] ) are independent of @xmath112 , we can assume @xmath113 without loss of generality in our proof .
assume the statement holds up to some @xmath114 .
( the constant @xmath115 will be determined in the course of the proof . ) _ case _ 1 : fix a @xmath116 for which @xmath117 for some @xmath118 to be specified later .
( in any case , we assume that @xmath119 . )
using an idea of @xcite , we may write for arbitrary @xmath120 , @xmath121 and @xmath122 the inequality @xmath123 let @xmath124 notice that @xmath125 set @xmath126 using the induction hypothesis and ( [ assumpgamma ] ) we obtain @xmath127\\ [ -8pt ] & & \quad\leq c\exp\biggl\ { -\frac{ct^{2}}{k+ b t^{\gamma}}\biggr\ } + a \exp\biggl\ { -%
\frac{a p^{2}t^{2}}{k + b p^{\gamma}t^{\gamma}}\biggr\ } + c\exp\biggl\ { -\frac{% c q^{2 } t^{2}}{n - k + b q^{\gamma}t^{\gamma}}\biggr\ } .\nonumber\end{aligned}\ ] ] notice that we chose @xmath128 to make the first and third terms in the right - hand side of ( mainineqgamma ) almost equal , and since by ( [ k ] ) @xmath129 the first term is greater than or equal to the third .
first we handle the second term in ( [ mainineqgamma ] ) , showing that for @xmath130 , @xmath131 for this we need to verify that for @xmath132 , @xmath133 which is equivalent to @xmath134 using that @xmath135 , \ ] ] it is enough to show @xmath136 note that if the coefficient of @xmath106 is positive , then we can choose @xmath137 in ( [ alpha ] ) small enough to make the above inequality hold , even if the coefficient of @xmath138 is negative .
so in order to guarantee ( [ eq1 ] ) ( at least for large @xmath106 ) we only have to choose the parameter @xmath139 so that @xmath140 which implies that @xmath141 holds and then select @xmath137 small enough .
next we treat the first and third terms in ( [ mainineqgamma ] ) . by the remark above ,
it is enough to handle the first term .
let us examine the ratio of @xmath142 and @xmath143 .
notice again that since @xmath144 , @xmath145 at this point we need that @xmath146 .
thus we choose @xmath137 small enough so that @xmath147 also , using @xmath148 , we get the bound @xmath149 which holds if @xmath106 is large enough .
therefore , we obtain for the ratio @xmath150 whenever @xmath151 , that is , @xmath152 . substituting back into ( [ mainineqgamma ] ) , for @xmath153 and @xmath148 we obtain @xmath154 where the last inequality holds for @xmath155 .
next assume that @xmath156 . in this case , choosing @xmath115 large enough , we can make the bound @xmath157 , namely @xmath158 if @xmath159 .
_ case _ 2 : now we must handle the case @xmath160 . here
we apply the inequality @xmath161 using assumption ( [ assumpgamma ] ) and the induction hypothesis , we have @xmath162 we will show that the right - hand side @xmath163 .
for this it is enough to prove @xmath164 first assume that @xmath165 . using the bound following from @xmath166 ,
we get @xmath167 we have that the right - hand side of ( [ less ] ) for @xmath168 is less than @xmath169 for @xmath115 large enough . for @xmath170
we have to use a different argument .
for @xmath171 large enough ( i.e. , for @xmath106 large enough , since @xmath166 ) we have @xmath172 we also have for @xmath173 , @xmath174 it is clear that since @xmath175 , for @xmath171 large enough , that is , for @xmath106 large enough , @xmath176 the proof is complete .
choose any @xmath177 and set @xmath178 for @xmath179 now , using inequality ( [ cineq ] ) , we get @xmath180 since as @xmath181 @xmath182 it is readily checked that for @xmath183 large enough so that @xmath184 @xmath185 and thus , since @xmath186 , we get by the borel cantelli lemma that with probability @xmath81 @xmath187 next we see that for all @xmath188 @xmath189 thus by ( [ aa ] ) , with probability @xmath190 @xmath191 hence , since @xmath177 can be chosen arbitrarily close to @xmath81 and @xmath66 arbitrarily close to @xmath68 , we have proved ( [ blil ] ) .
the authors are grateful to the referee for a number of penetrating comments and suggestions that greatly improved the paper .
kevei s research was partially supported by the analysis and stochastics research group of the hungarian academy of sciences and mason s by an nsf grant .
gin , e. and guillou , a. ( 2002 ) .
rates of strong uniform consistency for multivariate kernel density estimators . en lhonneur de j. bretagnolle , d. dacunha - castelle , i. ibragimov .
h. poincar probab . statist_.
* 38 * 907921 .
merlevde , f. , peligrad , m. and rio , e. ( 2009 ) .
bernstein inequality and moderate deviations under strong mixing conditions . in _
high dimensional probability v : the luminy volume _
( c. houdr , v. koltchinskii , d.m . mason and m. peligrad , eds . ) 273292 .
beachwood , oh : ims .
mricz , f.a .
exponential estimates for the maximum of partial sums .
i. sequences of rv s .
special issue dedicated to george alexits on the occasion of his 80th birthday .
_ acta math .
hungar . _ * 33 * 159167 .
statuleviius , v.a . and jakimaviius , d.a .
estimates for semiinvariants and centered moments of stochastic processes with mixing .
i. _ litovsk .
_ * 28 * 112129 ; translation in _ lithuanian math .
j_. * 28 * 6780 .
|
we show somewhat unexpectedly that whenever a general bernstein - type maximal inequality holds for partial sums of a sequence of random variables , a maximal form of the inequality is also valid . _ dedicated to the memory of sndor csrg _
|
the recent discovery of discrete , multiple stellar populations within several among the most massive globular clusters ( gc ) in the milky way ( e.g. , bedin et al .
2004 : piotto et al .
2007 ) has brought new interest and excitement on gc research .
further excitement was added by the realization that some of these stellar populations are selectively enriched in helium to very high values ( @xmath2 ) , without such enrichment being accompanied by a corresponding increase in the heavy element abundance of the expected size , if at all ( e.g. norris 2004 ; piotto et al .
2005 , 2007 ) .
thus , two main closely interlaced questions arise : how did such multiple populations form ? and , which kind of stars have selectively produced the fresh helium , without contributing much heavy elements ?
therefore , understanding the origin of multiple populations in gcs will need to make major steps towards a better knowledge of a variety of processes such as gc formation , star formation , as well as of some long standing issues in stellar evolution , e.g. , the asymptotic giant branch ( agb ) phase or the effect of rotation in massive stars .
moreover , it is quite possible that some of the massive gcs with multiple populations are the compact remnants of nucleated dwarf galaxies ( bekki & norris 2006 ) , an aspect that widens even further the interest for this subject .
all this together makes multiple populations in gcs an attractive , interdisciplinary subject of astrophysical investigation . in this paper
the main observational evidences are used to constrain the proposed possible scenarios for the origin of the multiple populations in gcs , and their associated chemical enrichment processes . in section 2 the main relevant observational facts
are briefly reviewed , and in section 3 agb stars and fast rotating massive stars are discussed as possible helium producers , along with population iii stars and the possible role of deep mixing during the red giant branch ( rgb ) phase of low mass stars . in section 4 various proposed scenarios for the origin of the multiple populations
are confronted with the relevant observational facts , arguing that only massive agb stars appear to remain viable as helium producers . a general discussion and the main conclusions are presented in section 5 .
the main observational facts concerning the evidence for multiple stellar populations in galactic gcs are briefly presented in this section , separately for photometric and spectroscopic observations .
further collective information and references can be found in a recent review on these topics by piotto ( 2008 ) .
it was recognized in the early seventies that stars in the gc @xmath0 cen span a wide range of metallicities ( cannon & stobie 1973 ; freeman & rodgers 1975 ) and since then it has been considered as a unique , exceptional cluster . over the last several years
there has been a surge of interest on this cluster , starting with the discovery that its _ broad _ rgb actually resolves into several distinct rgbs ( lee et al .
1999 ; pancino et al .
then came the discovery that also its main sequence ( ms ) splits into two parallel sequences , with the bluer one being more metal rich by nearly a factor of two , hence indicating a higher helium abundance ( bedin et al . 2004 ; norris 2004 ; piotto et al .
from the color - magnitude diagram ( cmd ) of the subgiant branch ( sgb ) region we now know that this cluster includes at least 4 , possibly 5 distinct stellar populations ( sollima et al 2005 ; lee et al .
2005 ; villanova et al . 2007 ) .
the majority of the cluster stars ( @xmath3 ) populate the _ red _ ms , have [ fe / h ] = 1.7 and are assumed to have primordial helium abundance , @xmath4 ; @xmath5 belong to the _ blue _ , helium enriched ms with [ fe / h ] = 1.4 and @xmath6 , which implies a huge helium enrichment ratio @xmath7 ( piotto et al . 2005 ) .
the residual @xmath8 of the stars belong to a metal rich component for which discrepant spectroscopic estimates exist , ranging from [ fe / h ] = 1.1 @xmath9 ( villanova et al .
2007 ) to [ fe / h ] = 0.6 ( pancino et al . 2002 ) .
part of the discrepancy may be due to the small number of stars in this group that have been observed at high resolution .
for these stars we have no direct hint on the helium abundance , but values as high as @xmath10 have been proposed ( sollima et al . 2005 ; lee et al . 2005 ) .
how these three ms components map into the many sgb , rgb and horizontal branch ( hb ) components of this cluster remains partly conjectural .
however , on the rgb there is well established evidence for sodium and aluminium being anticorrelated with oxygen and magnesium ( norris & da costa 1995 ; smith et al .
2000 ) , indicative that in a fraction of the stars material is present that was processed through hydrogen burning at high temperatures .
note that the @xmath11 overall fraction of the blue ms population takes into account the observed radial gradient in this fraction ( bellini et al . in preparation ) , the helium rich population being more centrally concentrated . with a mass of @xmath12 ,
@xmath0 cen is the most massive gc in the galaxy ( pryor & meylan 1993 ) . on this basis
, it is worth estimating the amount of fresh helium and iron that was produced and is now incorporated in the minority populations . given its mass ( @xmath13 ) and helium abundance , the intermediate metallicity , helium rich population includes @xmath14 of fresh helium ( i.e. , helium of stellar origin ) and @xmath15 of fresh iron , having adopted @xmath16 .
similarly , the most metal rich sub - population includes @xmath17 of fresh iron but we can not presently estimate its helium enrichment , if any .
based on the multimodal hb morphology of this cluster , dantona & caloi ( 2004 ) speculated that ngc 2808 harbours three main populations , each with a distinct helium abundance .
this has been nicely confirmed by the discovery of a triple ms in this cluster ( piotto et al .
2007 ) , with @xmath18 of the stars being assigned the primordial helium abundance ( @xmath4 ) , @xmath19 having helium enhanced to @xmath20 , and @xmath21 being up to @xmath22 .
the residual @xmath23 of the stars are likely to be binaries .
given the narrow rgb sequence , no iron abundance differences appear to be associated with these helium differences .
however , there appears to be a multimodal distribution of [ o / fe ] ratios among the rgb counterparts of the three ms populations ( carretta et al .
2006 ) , suggesting that most oxygen has been turned to nitrogen in the helium - enriched populations.thus , this cluster represents an even more extreme case as far as the helium enrichment parameter @xmath24 is concerned , as there is no detectable increase of the overall metallicity in the helium enriched populations .
the mass of this cluster is @xmath25 ( pryor & meylan 1993 ) , and therefore the intermediate helium - rich , and the very helium - rich populations contain @xmath26 and @xmath27 of fresh helium , respectively , with no appreciable extra iron .
no multiple main sequences have yet been detected in the cmd of this cluster , which clearly shows a double sgb , with the two components being nearly equally populated ( @xmath28 and @xmath29 ) , and separated by @xmath30 mag ( milone et al .
if interpreted in terms of age , this luminosity difference would imply an age difference of @xmath31 gyr .
alternatively , it has been proposed that the double sgb may be due to one of the two populations being enhanced in cno elements by a factor of @xmath32 relative to the other ( cassisi et al . 2008 ) .
the cmd of this cluster and of the superimposed core of the tidally disrupted sagittarius dwarf is extremely complex , with evidence of multiple ms turnoffs ( siegel et al .
2007 ) . however , there is no evidence of multiple populations within the cluster itself .
much observational and theoretical efforts have been dedicated to these two metal - rich bulge clusters , since the discovery that their hbs exhibit a remarkable blue extension ( rich et al .
unique among all gcs , they contain a number of rr lyraes with an average period as long as 0.75 days , making them a third oosterhoff type ( pritzl et al .
the similarity of their blue hbs with those of @xmath0 cen and ngc 2808 has been emphasized by busso et al .
( 2007 ) who further explore helium enrichment as the cause of their unusual hb ( as originally suggested by sweigart & catelan 1999 ) . based on cmds that include hst ultraviolet bands ( f225w and f336w ) , busso et al . advocate a very high helium enhancements ( up to @xmath33 in ngc 6388 and @xmath34 in ngc 6441 ) ) for @xmath19 of the stars in each cluster . a more moderate helium enrichment ( @xmath35 is suggested by yoon et al .
2007 ) , but their hb data do not include the very hot extension revealed by the uv observations .
both clusters have a mass of @xmath25 ( pryor & meylan 1993 ) , implying that the fresh helium content of the two clusters is @xmath36 and @xmath37 , respectively for ngc 6388 and ngc 6441 .
given their low galactic latitude , these gcs are affected by fairly high reddening ( hence differential reddening ) which has so far prevented checking for the presence of multiple mss and sgbs as would be expected if the interpretation of their hb morphology in terms of helium enhancement is correct . however , this situation may improve soon , and there is already an indication for a split in the sgb of ngc 6388 ( piotto 2008 ) .
departures from chemical homogeneity among stars within individual gcs are known since a long time ( see e.g. , the reviews by kraft 1979 , 1999 and gratton , sneden & carretta 2004 )
. generally referred to as _ abundance anomalies _ such departures include a variety of elements and molecules , such as cn and ch bimodality , the na - o , al - o , and al - mg anticorrelations , all indicative of contamination by materials having been exposed to hydrogen burning at high temperatures ( @xmath38 k ) .
some or all such anomalies are also exhibited by the massive clusters discussed in the previous section , but to some extent affect virtually all well studied gcs , irrespective of their mass .
for example , from the strength of the nh bands among rgb stars in ngc 6752 yong et al .
( 2008 ) infer than the nitrogen abundance among the sample stars spans almost two orders of magnitude , with no obvious bimodality . moreover , yong et al .
argue that a similar spread must exist in many other clusters , given that their whole rgb shows a spread in the nh - sensitive strmgren s @xmath39 index of similar size to that of ngc 6752 .
unfortunately , no similar @xmath39 data are presented for the clusters with multiple stellar populations discussed in the previous section . among them , ngc 2808 exhibits the canonical na - o anticorrelation , with three distinct peaks in oxygen abundance ( carretta et al .
2006 ) , that are likely to be associated with the three distinct populations revealed by the main sequence photometry ( piotto et al .
2007 ) . among the clusters mentioned in section 2.1 , ngc 6441 ( gratton et al . 2007 ) and ngc 6388 ( carretta et al .
2007 ) exhibit the na - o anticorrelation and so does ngc 1851 , which exhibits also al - o anticorrelation and variations of the s - process elements ( yong & grundahl 2008 ) . in @xmath0
cen star - to - star variations of cno elements exist as well , but their overall [ c+n+o / fe ] abundance ratio appear to be constant within a factor @xmath32 , as typical of all gcs ( e.g. , pilachowski et al . 1988 ; norris & da costa 1995 ; smith et al . 2000 , 2005 ) .
in particular , this holds for each of the various metallicity groups in @xmath0 cen , which also exhibit large dispersions in the al abundance ( johnson et al .
2008 ) the main observational constraints that will be used in the following to narrow down the options on the origin of the multiple stellar populations in gcs can be summarized as follows : all gcs with confirmed ( or highly probable ) multiple stellar populations ( @xmath0 cen , ngc 2808 , ngc 1851 , ngc 6388 , and ngc 6441 ) belong to the sample of the 10 most massive gcs in the galaxy ( with @xmath40 ) .
massive gcs with @xmath40 exist that do not show evidence for multiple mss and/or sgbs , nor multimodal hbs ( e.g. , 47 tuc , sirianni et al . 2005 ) .
some sub - populations can only be understood in terms of high helium content , up to @xmath41 or more .
multiple stellar populations within each gcs are characterized by discrete values of the helium and iron abundances , i.e. , there appears to be no composition spread within individual sub - populations as the width of the sequences on the cmds is consistent with being due only to known photometric errors .
clusters with helium - enriched multiple populations also tend to exhibit evidence for na and al being anticorrelated with o and mg , indicative of materials that have been exposed to hydrogen burning in a hot environment
. however , such variations appear to be virtually universal among gcs , no matter whether they exhibit multiple main sequences or not .
helium enrichment does not appear to be associated with an increase of [ c+n+o / fe ] .
a massive cluster ( m54 ) sits at the center of the core of the sagittarius dsph galaxy , and is embedded in its multiple stellar populations .
this offers a concrete example of a massive gc being the nucleus of a dwarf galaxy .
critical for understanding the origin of the multiple populations in gcs is the identification of the kind of stars responsible for the production of the excess helium now incorporated in some of these populations .
three kinds of stars have been discussed in the literature , namely : agb and super - agb stars , massive rotating stars , and population iii stars .
agb stars have long been considered for being responsible for at least some of the composition anomalies of gc stars ( dantona , gratton & chieffi 1983 ; renzini 1983 ; iben & renzini 1984 ) .
indeed , agb stars present two attractive characteristics , namely : 1 ) they eject large amounts of mass at low velocity ( @xmath42 km s@xmath43 ) which can then be retained within the potential well of gcs , and 2 ) the ejected materials can be highly processed through hydrogen burning at high temperature , hence being enriched in he and n , and presenting the na - o and al - o anticorrelations ( e.g. , renzini & voli 1981 ; dantona & ventura 2007 ; karakas & lattanzio 2007 ) .
however , three difficulties with the agb scenario have been pointed out : the helium abundance , the mass of the secondary populations relative to the primary one , and the constancy of [ c+n+o / fe ] ( e.g. , karakas et al .
2006 ; karakas & lattanzio 2007 ; choi & yi 2008 ) .
these difficulties are here addressed again . among agb stars ,
especially interesting are those in the mass range @xmath44 , because they experience the hot bottom burning ( hbb ) process that even in the presence of carbon third dredge - up ( 3du ) prevents the formation of carbon stars ( renzini & voli 1981 ) , and will produce the na - o and al - o anticorrelations . instead , in lower mass agb stars the c / o ratio largely exceeds unity , and especially so at low metallicities .
if stars were to form from these agb ejecta they would be carbon stars , whereas such stars are absent in the clusters with multiple populations .
moreover , @xmath44 stars experience the so - called second dredge - up ( 2du ) shortly before reaching the agb ( becker & iben 1979 ) , leading to a sizable helium enrichment in the whole stellar envelope .
besides the 2du , also the the 3du and the hbb may contribute to increase the helium abundance in the envelope of massive agb stars , see e.g. , fig .
11 in renzini & voli ( 1981 ) .
however , renzini & voli models assumed the validity of a universal core mass - luminosity relation for agb stars , whereas blcker & schnberner ( 1991 ) showed that this is no longer valid once the hbb process operates .
instead , in the presence of hbb the luminosity of agb stars increases dramatically , driving stars to very high ( superwind ) mass loss rates and leading to an earlier termination of the agb phase .
although this scenario is generally accepted , the precise duration of the agb phase , hence the extent to which the 3du and the hbb process operate in massive agb stars , all remain highly model dependent .
thus , a reasonable _ lower limit _ to the amount of helium enrichment is given by the 2du contribution alone , which is fairly well established .
this assumption is consistent with the blcker & schnberner ( 1991 ) result , which suggests a very prompt ejection of most of the envelope shortly after the onset of the hbb process . in practice
, there would be time for the hot cno processing of these elements originally present in the star , converting most of c and o into n ( renzini & voli 1981 ) , as well as for establishing the anticorrelations of na and al with o and mg .
these are indeed fairly rapid nuclear processes once the temperature at the base of the convective envelope is high enough .
moreover , the prompt ejection of the envelope drastically reduces the time spent on the thermally pulsing agb ( e.g. , over the early estimates of renzini & voli ) , suppressing along with it most of the 3du events , hence preventing an appreciable increase of the overall cno abundance in the envelope .
thus , it is quite plausible for agb stars with hbb to eject material in which helium is highly enriched , cno nuclei are globally not significantly enhanced , but have approached their nuclear equilibrium partition , and anticorrelations among other nuclei have been established by proton captures at high temperatures .
thus , in this scenario there is no significant increase of c+n+o due to the 3du . in summary
, it is assumed here that agb stars experiencing the hbb process ( @xmath45 ) 1 ) eject the whole envelope shortly after the onset of hbb , 2 ) the ejecta are enriched in helium solely by the 2du , 3 ) too few 3du events have time to take place , and no appreciable increase of the overall cno abundance occurs in the envelope , and 4 ) hbb is sufficiently effective to promptly establish the na - o and al - o anticorrelations . some observational evidences support these assumptions .
the mass distribution of white dwarfs terminates at @xmath46 ( bergeron , saffer & liebert 1992 ; bragaglia , renzini & bergeron 1995 ; koester et al .
2001 ) , indicating that the core mass of agb stars does not grow beyond this limit .
this also implies that there must be very little , if any , increase of the core mass during the evolution of the most massive agb stars , since the core mass of a @xmath47 star just after completion of the 2du is already @xmath48 ( becker & iben 1979 ) .
in addition , an agb calibration based on globular clusters in the magellanic clouds indicates that in clusters younger than @xmath49 myr there is negligible contribution of agb stars to the bolometric luminosity of the clusters ( maraston 2005 ) .
this implies a very short agb phase and a very small , if any , increase of the core mass during the agb phase of stars more massive than @xmath50 . assuming that the first stellar generation formed in a short , virtually instantaneous burst , @xmath51 to @xmath52 stars were shedding their envelope between @xmath53 and @xmath49 myr after the burst .
it is during this time interval that helium - enriched agb ejecta may have accumulated inside the cluster potential well , and new stars may have formed out of them .
the amount of fresh helium released by stars of initial mass @xmath54 assuming only the 2du contribution can be easily estimated from becker & iben ( 1979 ) , see also fig . 1 in renzini & voli ( 1981 ) , where the mass of dredged - up helium is @xmath55 for @xmath56 and increases linearly to @xmath57 for @xmath58 .
note that only 3/4 of the dredged - up helium is `` fresh '' , i.e. , has been synthesized within the star itself , whereas 1/4 is primordial , given that @xmath4 in the first stellar generation . therefore : @xmath59 which applies to a metal - poor population with @xmath60 . here
only stars with initial mass in the interval @xmath61 experience the 2du before the agb phase , then soon activate the hbb process and expel the envelope according to the scenario sketched above .
thus , convolving the mass of fresh helium with the initial mass function ( imf ) one can derive the total mass of fresh helium produced and expelled by these 3 to 8 @xmath62 stars per unit stellar mass in the whole population . for a salpeter imf ( slope @xmath63 ) for @xmath64 , and a flatter imf for the lower mass stars with @xmath65 for @xmath66 ,
one then derives : @xmath67 i.e. , the mass of fresh helium released is @xmath68 of the original mass of the parent stellar population . in a similar fashion
, one can estimate the helium abundance @xmath69 in the ejecta of the 3 to 8 @xmath62 stars , which ranges from the primordial value @xmath4 for @xmath56 to @xmath70 for @xmath71 . integrating over the same imf
, one gets that the average helium abundance of the ejecta from 3 to 8 @xmath62 stars is @xmath72 .
higher values could be obtained by restricting the integration over a narrower mass range , e.g. , 4 to 8 @xmath62 or 5 to 8 @xmath62 . for the minimum mass approaching 8 @xmath62
then @xmath73 tends to 0.35 , but the total amount of released fresh helium would vanish .
stars in the range @xmath74 ignite carbon non - explosively and may leave o - ne white dwarfs , or proceed to electron - capture / core - collapse supernovae . the former outcome results if the envelope is lost in a ( super)wind during helium - shell burning ( the super - agb phase ) .
conversely , a supernova explosion ends the life of the star if mass loss during the super - agb phase is insufficient to prevent the core from growing in mass until it finally collapses ( e.g. , nomoto 1984 ; ritossa , garcia - berro & iben 1996,1999 ; poelarends et al .
s - agb stars have also experienced the 2du and their envelope has been enriched in helium to @xmath6 , which makes them attractive helium contributors in the context of helium - rich populations in globular clusters ( pumo , dantona & ventura 2008 ) . assuming that all 8 to 10 @xmath62 s - agb stars leave o - ne white dwarfs
, one can estimate that the fresh helium mass produced by 3 to 10 @xmath62 stars would be @xmath75 higher than given by eq .
( 2 ) , i.e. , @xmath76 the accuracy of this theoretical estimate is difficult to assess
. the helium mass could actually be somewhat higher if the duration of the agb phase is long enough to allow the 3du and hbb processes to increase the envelope helium beyond the value reached after the 2du , but it could be lower if the mass range of the useful s - agb stars is narrower than adopted here .
( note that in both cases the main uncertainty comes from what one is willing to assume for the mass loss during the agb / s - agb phases . )
still , one can regard this estimate as quite reasonable , given our current understanding of agb / s - agb evolution .
as far as the average helium abundance is concerned , from the inclusion of s - agb stars one predicts a modest increase from @xmath72 to @xmath77 , or a little higher if considering a minimum contributing mass somewhat in excess of @xmath78 . in any event
, @xmath79 can be regarded as the upper limit given the assumed agb / s - agb evolution .
finally , it is worth noting that stars below @xmath1 do not experience the 2du and the hbb processes , spend a long time on the agb in its thermally - pulsing phase , and experience repeated 3du episodes which increase both helium and carbon in their envelopes .
the fact that helium is accompanied by carbon enhancement makes these lower mass agb stars less attractive helium producers , because the helium rich populations in gcs do not appear to be enriched in carbon ( piotto et al .
thus , the interesting mass range is @xmath51 to @xmath80 . in summary
, it is assumed that metal poor ( @xmath81 ) agb stars more massive than @xmath1 and s - agb stars experience the 2du and the hbb process , and leave enough for the hbb process to convert most of the original carbon and oxygen into nitrogen , but not enough to experience a sufficient number of 3du episodes to significantly increase the overall cno abundance in the envelope . in this respect ,
the schematic agb evolution adopted here differs from agb models existing in the literature ( e.g. , renzini & voli 1981 ; groenewegen & de jong 1993 ; herwig 2004 ; izzard et al .
2004 ; ventura & dantona 2005 , 2008 ; marigo & girardi 2007 ) .
massive stars also produce sizable amounts of fresh helium , especially during their wolf - rayet phase .
however , they also produce metals in large quantity , whereas the helium rich populations in gcs are very modestly enriched in metals ( @xmath0 cen ) , or not at all ( ngc 2808 ) . in the attempt to overcome
this difficulty , maeder & meynet ( 2006 ) have proposed fast - rotating massive stars as potential helium producers in young gcs , a scenario further developed by decressin et al .
( 2007 ) , decressin , charbonnel & meynet ( 2007 ) and meynet , decressin & charbonnel ( 2008 ) .
massive rotating stars would harbor meridional circulations bringing to the surface products of hydrogen burning ( i.e. , helium ) , while losing ( helium enriched ) mass in three distinct and physically separated modes : 1 ) a slow outflowing equatorial disk whose helium abundance increases as evolution proceeds , 2 ) a regular , fast , radiatively driven wind also enriched in helium in the directions unimpeded by the disk , and 3 ) a final core - collapse supernova explosion .
only the slow outflowing disk is considered of interest for the production of helium - enriched stars , because both radiative winds and supernova ejecta run at thousands of km / s , and would not be retained inside the relatively shallow potential well of the proto - cluster .
while physically plausible , this scenario suffers from the difficulty of predicting with any degree of confidence the efficiency of meridional circulations to mix helium into the stellar envelope , and the rate of mass loss via the outflowing disk , none of which can be estimated from first principles .
moreover , star formation would have to be confined within the outflowing disks around individual stars , before such disks are destroyed by the fast winds and supernova ejecta from nearby cluster stars , and mixed with them ( meynet et al .
2008 ) . in an early attempt to account for some gc abundance anomalies ,
sweigart & mengel ( 1979 ) proposed that in low mass upper rgb stars ( @xmath82 ) mixing could extend below the formal boundary of the convective envelope , and reach well into the hydrogen burning shell .
thus , materials processed by hydrogen - burning reactions within the shell could be brought to the surface , changing the c : n : o proportions , and possibly establishing some of the abundance ( anti)correlations typical of gc stars . moreover , along with processed cno elements , some helium enrichment could also take place , hence affecting the subsequent hb evolution ( sweigart 1997 ; sweigart & catelan 1998 ; moehler & sweigart 2006 ) .
observations of upper rgb stars in several gcs suggests indeed that the sweigart - mengel process may be at work in these stars , given that the @xmath83c/@xmath84c ratio has reached close to its nuclear equilibrium value ( @xmath85 ) in virtually all of them ( recio - blanco & de laverny 2007 ) .
however , several abundance anomalies extend to much lower luminosities and down to the main sequence ( e.g. gratton et al .
2004 ) , which means that rgb self - enrichment can not be the only process at work .
in particular , it can not account for the helium - enriched ms stars , unless one is willing to consider the possibility of forming the subsequent stellar generation out of the ejecta from low - mass rgb stars .
there are several difficulties with this option , such as the small mass lost by stars that may experience the sweigart - mengel process ( i.e. , those in the range @xmath86 to @xmath87 ) , relative to agb / s - agb stars , or the very long ( several gyr ) time required to accumulate any sizable amount mass before being suddenly turned into stars .
this very long accumulation time would imply the secondary populations to be several gyr younger than the first generation . according to villanova et al .
( 2007 ) there is actually a hint for the helium rich population in @xmath0 cen being a few gyr younger , but other interpretations of the data exist that do not require such large age differences ( sollima et al .
2005 ; lee et al .
moreover , a several gyr long accumulation time looks quite unlikely , given the possible interaction of the cluster with the galactic environment ( e.g. disk crossing , tidal stripping , etc . ) .
perhaps more fundamentally , metal poor @xmath88 stars become carbon stars on the agb , the phase during which most of their mass loss takes place , and therefore they are not suited to provide raw material with the proper chemical composition for the production of secondary populations . for all these reasons rgb stars are considered less likely helium producer candidates .
still , rgb self enrichment may add further variance on top of other processes .
finally , in order to achieve very high @xmath24 values stochastic contamination by helium produced by population iii stars was suggested by choi & yi ( 2007 ) in a scenario in which population iii and population ii star formation temporally overlap . here
helium would be produced by very massive pop .
iii stars , and metals by massive pop .
ii stars , and mixing their products in various proportions one could achieve high values of @xmath89 .
although this process may take place in nature , it does not appear to work for explaining the properties of multiple stellar populations in globular clusters .
for example , in the case of ngc 2808 we have three distinct values of the helium abundance , with the same metallicity , a pattern that can not be reproduced by mixing gas that formed the dominant cluster population with helium - rich gas from pop .
iii stars .
enrichment in helium would indeed be accompanied by dilution of metals , and the helium rich population would be metal poor , contrary to the case of both @xmath0 cen and ngc 2808 .
a variant of this scenario was proposed by chuzhoy ( 2006 ) , with population iii stars being pre - enriched in helium thanks to gravitational settlement of this element within dark matter halos in the early universe .
while this mechanism may produce even higher helium abundances in the ejecta of massive population iii stars , it is prone to the same difficulties of the choi & yi scenario if intended to account for the helium rich populations in gcs .
a most stringent constraint on scenarios for the origin of multiple populations is their very nature , i.e. , being indeed , multiple , discrete populations each characterized by a specific helium abundance , as most clearly evident in the case of @xmath0 cen and ngc 2808 .
this immediately implies that helium enrichment of the interstellar medium ( ism ) and star formation for the second ( and third ) stellar generation were not concomitant processes , but instead took place sequentially .
helium - rich material was accumulated in the ism ( and mixed therein ) for a sufficiently long time until suddenly a burst turned a major fraction of the ism into stars .
in fact , a continuous star formation process proceeding along with the ism helium enrichment would have resulted in a continuous distribution of helium abundances in the newly formed stars , hence in a broadening of the main sequence rather than in well separated sequences as actually observed . in this section one
tests various proposed scenarios against this important constraint .
any gas matter lost by the first stellar generation ( e.g. , by its agb stars ) was necessarily less massive than the parent cluster , and hence had a lower mass density compared to the stellar component if spatially distributed in roughly the same way .
a simple calculation shows that under such circumstances any stellar mass size portion of the ism already included several lower main sequence stars . at first sight accretion onto these pre - existing seeds
would seem more likely than starting new stars from scratch .
hence , already a long time ago it was favored as a likely mean to produce chemical anomalies in gcs ( e.g. , dantona , gratton & chieffi 1983 ; renzini 1983 ; iben & renzini 1984 ) , and has been considered even recently in the context of the helium rich populations ( tsujimoto , shigeyama & suda 2007 ; newsham & terndrup 2007 ) . however , accretion depends on mass , velocity and orbit of each star inside the cluster , which would have resulted in a broad range of accreted masses from the ism whose helium abundance was secularly evolving .
after rayleigh - taylor mixing of accreted matter with the underlying layers of lower molecular weight , stars would now show a range of helium abundances .
thus , accretion would inevitably result in a broad , continuous distribution of stellar helium abundances , contrary to the required multiple discrete values . in spite of its attractiveness
, accretion must therefore be rejected as the primary process having produced the helium - enriched populations , at least in the two clusters with clearly defined multiple main sequences .
this leaves multiple star formation episodes as the only mechanism able to produce successive stellar generations , each with a uniform helium ( and metal ) abundance .
this means that helium - enriched gas had to accumulate in the potential well of the cluster , without experiencing any star formation , until something suddenly triggered the burst of star formation .
yet , it is unlikely that the efficiency of converting ism gas into stars was near unity , and the lower this efficiency , the more massive had to be the first stellar generation in order to account for the mass of the second generation and its helium content .
any gas not converted into stars was soon lost from the proto - cluster , blown in a wind powered by high velocity stellar winds from massive stars and supernova explosions .
it is reasonable to assume that the first cluster stellar generation arose from a burst of star formation , of duration shorter than the lifetime of the most massive stars ( i.e. , @xmath90 myr ) .
then the subsequent 2030 myr were dominated by massive stars , whose winds and supernovae cleared the protocluster of any residual gas .
after the last core - collapse supernova , i.e. , 2030 myr since the beginning , conditions were finally established favoring the low - velocity winds from s - agb / agb stars to accumulate . at the beginning of the accumulation
the helium abundance of such material was that pertaining to the ejected envelopes of @xmath91 stars , or @xmath92 according to the estimate presented in section 3.1 .
later , as agb stars of lower and lower mass were ejecting their envelope , contributing material less enriched in helium , the helium abundance in the ism kept steadily decreasing , down to @xmath93 some 300 myr after start .
subsequently , as the agb started to be populated by @xmath94 stars , addition of fresh helium from 3du replaced that from 2du , but along with it also carbon and s - process elements began to accumulate . clearly , the time interval between 2030 myr and @xmath95 myr is the most propitious epoch for producing subsequent stellar generations with an helium abundance and overall composition close to that demanded by the observations .
more demanding is , however , the requirement of producing a second or a third generation of the observed mass .
for example , in the case of @xmath0 cen , according to eq .
( 3 ) its present first generation of @xmath96 produced only @xmath97 of fresh helium , even allowing for a full s - agb contribution , a factor of @xmath98 less than the observational requirement estimated in section 2.1.1 .
we encounter here the main difficulty of the agb scenario , as already pointed out by several authors ( e.g. , karakas et al .
2006 ; bekki & norris 2006 ; karakas & lattanzio 2007 ) , i.e. , requiring a first generation of stars much more massive than that still harbored by the cluster .
this difficulty is further exacerbated if the actual efficiency of star formation is less than unity .
however , the helium abundance in agb / s - agb stars @xmath99 is not in sharp contrast with the observational requirements from the two cases with well established multiple main sequences ( @xmath0 cen and ngc 2808 ) , where in both cases @xmath100 .
thus , agb / s - agb stars remain viable candidates , though we need much more of them than the agb / s - agb share of the first generation we now see in these clusters .
one may think that one way of having more agb / s - agb stars is by making recourse to a flat imf .
however , this may work only if the imf of the subsequent generations is different from that of the first one , and in particular much steeper than it : a very contrived scenario .
otherwise , if the imfs are the same , with a flat imf one gets more helium from the first generation , but by as much increases the amount of fresh helium demanded by the second generation , and the discrepancy by a factor of @xmath98 remains .
thus , a flat imf does not solve the problem .
it is very difficult to prove or disprove those models of fast rotating massive stars ( frms ) in which helium is mixed in the stellar envelopes by meridional circulations , and slowly lost in an outflowing disk .
thus , i take the frms scenario as described in decressin et al . (
2007 ) at face value .
one problem with this scenario is that it assumes that such disks survive long enough to produce new stars , in spite of their impervious environment in which they are bombarded from all directions by fast stellar winds and supernova explosions .
but even admitting that disks manage to deliver new stars , such stars will reflect the helium abundance of the disks they are born from , which varies from one massive star to another , and for any given stars varies as a function of its evolutionary stage .
thus , stars born out of such disks will inevitably show a spread in helium abundances , and one would have broad gc main sequences rather than multiple ones as demanded by the observations . the existence of multiple , discrete stellar populations in @xmath0 cen and ngc 2808 rules out this frms scenario as a viable one to explain the helium rich populations .
nevertheless , it is possible that meridional circulations are at work in massive stars , and bring fresh helium to the surface .
if so , some meridional mixing and helium enhancement may not be confined to the very massive stars exploding as supernovae
. some helium enrichment might also take place during the main sequence phase of stars less massive than @xmath91 , which will later deliver such helium during their agb / s - agb phase .
therefore , if meridional mixing of helium exists in @xmath101 main sequence stars , it would alleviate the difficulty for the agb / s - agb scenario outlined in the previous subsection , in particular concerning the helium abundance , yet by an amount that is hard to guess theoretically .
spectroscopic observations of @xmath101 stars appear to be the only way to assess whether helium enrichment does indeed take place , either directly from helium line strengths in hot stars , or indirectly from c : n : o ratios indicative of deep mixing .
contrary to the case of ngc 2808 , in @xmath0 cen the helium - enriched population identified by the blue main sequence is also enriched in iron , and contains some @xmath102 of fresh iron that was not initially present in the first stellar population ( cf .
section 2.1.1 ) .
if coming from relatively prompt type ia supernova events , each contributing @xmath103 of iron ( e.g. , iwamoto et al .
1999 ) , at least @xmath104 type ia supernovae ( snia ) from the first stellar generation had to explode within the helium enriched ism , and do so within the first @xmath105 yrs .
this looks quite plausible given the wide variety of distributions of snia delay times that theoretical models can generate ( e.g. greggio 2005 ) .
however , if the excess iron were produced by snia s , then the secondary population would have lower @xmath106-element to iron ratios , being selectively enriched only in iron .
this is at variance with the observed [ ca / fe ] , [ mg / fe ] and [ si / fe ] ratios in @xmath0 cen rgb stars , which show no dependence on the iron abundance ( norris & da costa 1995 ; smith et al .
2000 ; pancino et al .
thus , this excludes snia s for being responsible for the iron enrichment , and favours core collapse supernovae ( ccsn ) , which along with iron produce also @xmath106 elements . on average
, each ccsn produces @xmath107 of iron ( hamuy 2003 ) , roughly 10 times less than each snia .
thus , a few hundred ccsne are needed to produce the 25 @xmath62 of iron in the helium - rich population of @xmath0 cen . with the adopted imf ,
one ccsn is produced every @xmath108 of gas turned into stars , and therefore with its @xmath109 the first stellar generation in @xmath0 cen has produced @xmath110 ccsne . if the progenitor was at least 10 times more massive than the present cluster ( see below ) , then over @xmath111 ccsne had been produced . thus , it is sufficient that @xmath112 of their ejecta were trapped inside the protocluster while mass lost by s - agb / agb stars had already started to accumulate .
a very small time overlap between ccsn events and the appearance of s - agb / agb stars is necessary for this to happen , consistent with the short timescale ( @xmath90 myr ) postulated for the formation of the first stellar generation ( cf .
section 4.2.1 ) .
note that these supernova explosions should have avoided to trigger any major star formation , otherwise a continuous distribution of helium and iron abundances would have resulted .
what triggered the major star formation burst leading to the second generation remains unidentified .
no attempt is made here to speculate on the full star formation history in @xmath0 cen , which is much more complex given the identification of its 5 sub - populations ( sollima et al .
2005 ; lee et al .
2005 ; villanova et al .
in the previous sections it has been argued that only the agb / s - agb scenario remains viable to account for the helium - enriched sub - populations .
still , with a serious difficulty to overcome , plus some minor ones .
if our current understanding of the helium enrichment in intermediate mass stars is not grossly incorrect , then the mass of the first stellar population in a cluster such as @xmath0 cen had to be several times larger than the present mass of the cluster .
following bekki & norris ( 2006 ) this difficulty can be solved if clusters such as @xmath0 cen and ngc 2808 are the remnant nucleus of a nucleated dwarf galaxy that was torn apart by the tidal field of the galaxy .
hence the parent first population providing the necessary raw material for the successive generation(s ) would have been much more massive than these clusters are today . some circumstantial evidence for this now widely entertained scenario is the finding that m54 , one of the most massive gcs in the galaxy , is indeed associated with the sagittarius dwarf , albeit there is no evidence for multiple populations _ within _ this cluster ( siegel et al .
2007 ) . the nucleated dwarf ngc 205
may be another example relevant to this scenario : with its tidal stream towards m31 ( mcconnachie et al .
2004 ) it may represent an early stage of the process suggested by bekki & norris .
its nucleus is dominated by old stellar populations , and yet it is quite bright in the wfpc2 ultraviolet f225w and f185w passbands ( cappellari et al .
if the uv light comes from an extended hb ( such as that of @xmath0 cen and ngc 2808 ) it may also harbor helium - enriched populations , and would represent a possible testbed for the nucleated dwarf scenario of bekki & norris ( 2006 ) .
of course , for the nucleated dwarf scenario to work , successive stellar populations must have a much lower probability of being tidally stripped compared to the first population , otherwise the mass discrepancy would remain .
this can only be achieved if the successive starbursts are far more centrally concentrated compared to the first one , i.e. , the agb / s - agb ejecta from the first generation should collapse to the very bottom of the potential well before leading to star formation . in this connection , it is quite reassuring that the helium - enriched main sequence stars in @xmath0 cen are indeed markedly more centrally concentrated than the others ( cf .
section 2.1.1 ) . besides reproducing the required helium enhancement , a successful theory for
the origin of multiple stellar populations in massive gcs should also account for the observed abundance and distribution of cno and other intermediate mass elements .
several authors ( e.g. , karakas et al .
2006 ; romano et al . 2007
; choi & yi 2008 ) have shown that yields of existing agb models fail to satisfy this constraint , as along with helium stellar ejecta would be enriched also in carbon from the 3du .
the question is therefore as to whether one should conclude that the fresh helium does not originate from agb stars , or that the used theoretical yields are not correct .
given the current uncertainties affecting the massive agb models ( especially due to the treatment of mass loss and convective overshooting ) it seems worth keeping a pragmatic approach , hence taking existing agb models with special caution .
the schematic agb evolution described in section 3.1 is quite plausible given our ignorance of mass loss in bright agb / s - agb stars with hbb , and does not conflict with existing observations .
actually , it allows keeping agb / s - agb stars as viable helium producers for the multiple stellar generations in globular clusters . for the present scenario to work it
is essential that the agb ejecta accumulate for a fairly long time ( @xmath105 yr ) without any significant star formation before suddenly a major fraction of the interstellar medium is turned into stars by a burst .
this may not be such an _ ad hoc _ assumption , given that star formation in sporadic bursts appears to be the norm for dwarf galaxies ( gerola , sneden & sulman 1980 ) , as also demonstrated by the discrete multiple generations in dwarfs such as carina ( e.g. , monelli et al . 2003 ) .
in the case of ngc 2808 there had to be a second and a third stellar generation .
if the scenario presented in this paper is basically correct , then the most helium rich secondary sub - population would have been the first to form out of the most massive agb / s - agb ejecta , which are the most helium rich . then the ism was replenished again by the ejecta of the less massive agb stars from the first generation , plus perhaps a contribution by the secondary generation .
if so , then the generation identified by the _ middle _ of the three main sequences in this cluster was the last to form .
worth briefly mentioning are those observational studies that may help testing the plausibility of the agb / s - agb scenario proposed in this paper .
some of these tests concern the adopted agb / s - agb evolution and nuclear yields , in particular concerning @xmath51 to @xmath80 stars
. it would be interesting to test whether the surface helium and nitrogen abundance are enhanced in @xmath113 to @xmath80 main sequence stars , possibly by the meridional circulation process advocated by maeder & meynet ( 2006 ) , albeit high rotational velocities may hamper accurate abundance determinations , and massive stars a metal poor as stars in @xmath0 cen and ngc 2808 are not within reach .
direct observations of bolometrically very luminous agb / s - agb stars in the magellanic clouds with very high mass loss rates should help understanding the crucial , final evolutionary stages of intermediate mass stars .
moreover , high resolution spectroscopy of very large samples ( several hundreds ) of stars in the various sub - populations in globular clusters should help identify unequivocal chemical signatures of the _ donors _ of the materials out of which these sub - populations have formed .
these kind of studies are now possible thanks to the high multiplex multiobject spectrographs at 810 m class telescopes , such as e.g. , flames at the vlt ( sollima et al .
2005 ; carretta et al .
2006 ; villanova et al .
finally , the multifrequency study of nucleated dwarfs in and around the local group may help testing whether the most massive globular clusters may have originated from the tidal stripping of these objects . in conclusion , excluding scenarios that _ qualitatively _ conflict with observations , such as accretion , fast rotating massive stars or population iii stars ,
turn out to be much easier than proving others that qualitatively appear to work , but may have _ quantitative _ difficulties .
this is the case for the agb / s - agb scenario advocated in this paper , where the predicted helium abundance in the secondary populations admittedly falls a little short of the highest values suggested in the literature ( @xmath114 ) . in this respect
, one should consider than helium abundance estimates are affected by uncertainties that must be of the order of a few 0.01 , hence no macroscopic discrepancy appear to exist .
still , it would help if , besides the 2du , other processes contribute a little additional fresh helium in @xmath51 to @xmath115 stars .
additional helium may come from the hbb process operating during the agb / s - agb phase , and/or from meridional circulations during the main sequence phase of the progenitors of agb / s - agb stars .
i would like to thank giampaolo piotto for numerous stimulating discussions on the multiple populations in globular clusters and for a critical reading of the manuscript .
useful suggestions on the metal enrichment by supernovae are acknowledged from laura greggio .
finally , i would like to thank the anonymous referee for his / her many constructive comments .
becker s.a . &
iben i.jr . 1979 , apj , 232 , 831 bedin l.r . , piotto g. , anderson j. , cassisi s. , king i.r . , momany y. , carraro g. 2004 , apj , 605 , l125 bekki , k. & norris , j.e . 2006 , apj , 637 , l109 bergeron p , saffer r.a . , liebert j. 1992 , apj , 394 , 228 blcker t. , schnberner d. 1991 , a&a , 244 , l43 bragaglia a. , renzini a. , bergeron p. 1995 , apj , 443 , 735 busso g. , cassisi s. , piotto g. , castellani m. , romaniello m. , catelan m. , djorgovski s.g . , recio - blanco a. , et al .
2007 , a&a , 474 , 105 cannon , r.d . & stobie , r.s . 1973 , mnras , 162 , 207 cappellari m. , bertola f. , burstein d. , buson l.m . , greggio l. , renzini a. 1999 , apj , 515 , l17 carretta e. , bragaglia a. , gratton r.g . , leone f. , recio - blanco , a. , lucatello s. 2006 , a&a , 450 , 523 carretta e. , bragaglia , a. , gratton , r.g . , momany y. , recio - blanco a. , cassisi s. , franois p. , james g. , et al .
2007 , a&a , 464 , 967 cassisi s. , salaris m. , pietrinferni a. , piotto g. , milone a.p .
, bedin l.r . , anderson j. 2008 , apj , 672 , l115 choi e. , yi s.k .
2007 , mnras , 375 , l1 choi e. , yi s.k .
2008 , arxiv:0804.1598 chuzhoy l. 2006 , mnras , 369 , l52
dantona f. , caloi v. 2004 , apj , 611 , 871 dantona f. , gratton r. , chieffi a. 1983 , mem .
s. a. it . , 54 , 173 dantona f. , ventura p. 2007
, mnras , 379 , 1431 decressin t. , charbonnel c. , meynet g. 2007 , a&a , 475 , 859 decressin t. , meynet g. , charbonnel c. , prantzos n. , ekstrm s. 2007 , a&a , 464 , 1029 freeman k.c . , rodgers , a.w .
1975 , apj , 201 , l71 gerola h. , sneden p.e . , schulman l.s . 1980 , apj , 242 , 517 gratton r. , lucatello s. , bragaglia a. , carretta e. , cassisi s. , momany y. , pancino , e. , valenti , e. , et al .
2007 , a&a , 464 , 953 gratton r. sneden c. , carretta e. 2004 , ara&a , 42 , 385 greggio l. 2005 , a&a , 441 , 1055 groenewegen m.a.t . , de jong t. 1993 , a&a , 267 , 410 hamuy m. 2003 , apj , 582 , 905 herwig f. 2004 , apjs , 155 , 651 iben , i.jr . & renzini , a. 1984 , physics reports , 105 , no . 6 iwamoto k. , brachwitz f. , nomoto , k. , kishimoto n. , umeda h. , hix w.r .
, thielemann f .- k .
1999 , apjs , 125 , 439 izzard r.g .
, tout c.a .
, karakas a.i . , pols o.r .
2004 , mnras , 350 , 407 johnson c.i . , pilachowski c , a . , simmerer j. , schwenk d. 2008 , apj , 681 , 1505 karakas a. , fenner y. , sills a. , campbell s.w . ,
lattanzio j.c .
2006 , apj , 652 , 1240 karakas a. , lattanzio , j.c .
2007 , publ .
astr . soc .
australia , 24 , 103 koester d. , napiwotzki r. , christlieb n. , drechsel h. , hagen h .- j .
, heber u. , homeier d. , karl c. , et al .
2001 , a&a .
378 , 556 kraft r.p .
1979 , ara&a , 17 , 309 kraft r.p .
1994 , pasp , 106 , 553 lee y .- w . ,
m . , sohn y .- j . , rey s .- c . , lee h .- c . ,
walker a.r .
1999 , nature , 402 , 55 lee y .- w . ,
joo s .- j . , han s .-
i . , chung c. , ree c.h .
, sohn y .- j . , kim y .- c . ,
yoon , s .- j .
, et al . 2005 , apj , 621 , l57 maeder a. , meynet , g. 2006 , a&a , 448 , l37 maraston , c. 2005 , mnras , 362 , 799 marigo p. , girardi l.2007 , a&a , 469 , 239 mcconnachie a.w . , irwin m.j . , lewis g.f . , ibata r.a . , chapman s.c . ,
ferguson a.m.n . ,
tanvir n.r .
2004 , mnras , 351 , l94 meynet g. , decressin t. , charbonnel c. 2008 , mem.s .
a. it . , 79 , 584 milone a.p . ,
bedin l.r . , piotto g. , anderson j. , king i.r . , sarajedini a. , dotter a. , chaboyer b. , et al .
2008 , apj , 673 , 241 moehler s. , sweigart a.v .
2006 , a&a , 455 , 943 monelli m. , pulone l. , corsi c.e . , castellani m. , bono g. , walker a.r .
, brocato e. , buonanno r. , et al .
2003 , aj , 126 , 218 newsham g. , terndrup d.m .
2007 , apj , 664 , 332 nomoto k. 1984 , apj , 277 , 791 norris j.e .
2004 , apj , 612 , l25 norris j.e . , da costa g.s .
1995 , apj , 447 , 680 pancino e. , ferraro f.r .
, bellazzini m. , piotto g. , zoccali m. 2000 , apj , 534 , l83 pancino e. , pasquini l. , hill v. , ferraro f.r . , bellazzini m. 2002 , apj , 568 , l101 piotto , g. 2008 , mem .
s. a. it . , 79 , 334 piotto , g. , et al .
2005 , apj , 621 , 777 piotto g. , bedin l.r . , anderson j. , king i.r . , cassisi s. , milone a.p . , villanova s. , pietrinferni a. , et al .
2007 , apj , 661 , l53 poelarends a.j.t . , herwig , f. , langer , n. , heger , a.2008 , apj , 675 , 614 pritzl b. , smith h.a . , catelan m. , sweigart a.v . 2000 , apj , 530 , l41 pryor c. , meylan g. 1993 , in structure and dynamica of globular clusters , ed .
djorgovski & g. meylan , asp conf .
50 , 357 pumo m.l . , dantona f. , ventura p. 2008
, apj , 672 , l25 recio - blanco a. , de laverny p. 2007 , a&a , 461 , l13 renzini a. 1983 , mem . s. a. it . , 54 , 335 renzini a. & voli , m. 1981 , a&a , 94 , 175 rich r.m . , sosin c. , djorgovski s.g . , piotto g. , king i.r . , renzini a. , phinney e.s . , et al
1997 , apj , 484 , l25 ritossa c. , garcia - berro e. , iben i.jr .
1996 , apj , 460 , 489 ritossa c. , garcia - berro e. , iben i.jr . 1999 , apj , 515 , 381 romano d. , matteucci f. , tosi m. pancino e. , bellazzini m. , ferraro f.r . , limongi m. , sollima a. 2007 , mnras , 376 , 405 siegel m.h . , dotter a. , majewski s.r .
, sarajedini a. , chaboyer b. , nidever d.l . , anderson j. , marn - franch a. , et al .
2007 , apj , 667 , l57 sirianni m. , jee m.j .
, bentez n. , blakeslee j.p . ,
martel a.r .
, meurer g. , clampin m. , de marchi g. , et al .
2005 , pasp , 117 , 1049 smith v.v .
, suntzeff n.b . ,
cunha k. , gallino r. , busso m. , lambert d.l . , straniero o. 2000 , aj , 119 , 1239 smith v.v . , cunha k. , ivans i.i . , lattanzio j.c .
, campbell s. , hinkle k. 2005 , apj , 633 , 392 sollima a. , pancino e. , ferraro f.r . , bellazzini m. , straniero o. , pasquini l. 2005 , apj , 634 , 332 sweigart a. v. 1997 , apj 474 , l23 sweigart a. v. , catelan m. 1998 , apj , 501 , l63 sweigart a.v .
, mengel j.g . 1979 , apj , 229 , 624 tsujimoto t. , shigeyama t. , suda y. 2007 , apj , 654 , l139 ventura p. , dantona f. 2005 , a&a , 431 , 279 ventura p. , dantona f. 2008 , a&a , 479 , 805 villanova s. , piotto g. , king i.r . , anderson j. , bedin l.r . , gratton r.g . , cassisi s. , momany y. , et al .
2007 , apj , 663 , 296 yoon s .- j . ,
podsiadlowski ph . ,
rosswog s. 2007 , mnras , 380 , 933 yong d. , grundahl f. 2008 , apj , 672 , l29 yong d. , grundahl f. , johnson j.a .
, asplund m. 2008 , arxiv:0806.0187
|
the various scenarios proposed for the origin of the multiple , helium - enriched populations in massive globular clusters are critically compared to the relevant constraining observations . among accretion of helium - rich material by pre - existing stars , star formation out of ejecta from massive agb stars or from fast rotating massive stars , and pollution by population iii stars , only the agb option appears to be viable .
accretion or star formation out of outflowing disks would result in a spread of helium abundances , thus failing to produce the distinct , chemically homogeneous sub - populations such as those in the clusters @xmath0 cen and ngc 2808 .
pollution by population iii stars would fail to produce sub - populations selectively enriched in helium , but maintaining the same abundance of heavy elements .
still , it is argued that for the agb option to work two conditions should be satisfied : i ) agb stars experiencing the hot bottom burning process ( i.e. , those more massive than @xmath1 ) should rapidly eject their envelope upon arrival on the agb , thus experiencing just a few third dredge - up episodes , and ii ) clusters with multiple , helium enriched populations should be the remnants of much more massive systems , such as nucleated dwarf galaxies , as indeed widely assumed . [ firstpage ] _ ( galaxy : ) _ globular clusters : general _ ( galaxy : ) _ * globular clusters : individual : @xmath0 cen , ngc 1851 , ngc 2808 , ngc 6388 , ngc 6441 , m54 * stars : agb and post - agb
|
networks are widely prevalent in all spheres of life @xcite . a network of acquaintances is the simplest example one can think of . social , economic and political networks of various kinds are part of human society .
the internet , a network of information resources , plays a vital role in the gathering , sharing and transmission of information .
a network consists of nodes connected by links .
figure 1 shows the example of a network in which the solid circles denote the nodes and the solid lines the links .
some examples of real life networks are as follows : in a network describing an electrical power grid , the generators , transformers and substations are the nodes and the high - voltage transmission lines connecting them the links . in the world wide web ( www )
, the documents / pages constitute the nodes .
these are connected to other documents / pages through links . in a collaboration graph of movie actors ,
the nodes represent the actors .
two actors are connected by a link if they appear in the same movie . in a citation network ,
the nodes are the papers published in refereed journals .
a paper is linked to all the other papers it cites .
cellular processes are controlled by various types of biochemical networks . a metabolic network @xcite controls the processes which generate mass and energy from nutritive matter .
the nodes in such a network are the substrates such as atp , adp and @xmath0 .
two substrates are connected by a link if both of them participate in the same biochemical reaction .
traditional cell biology assigns specific functional roles to individual proteins , such as catalysts , signalling molecules and constituents of cellular matter . in the post - genomic era
, there is an increasing emphasis on understanding the functions of proteins as parts of an interacting network and also on the collective , emergent properties of the network . in a protein - protein interaction network @xcite ,
the nodes represent the proteins .
a link exists between two nodes if the corresponding proteins have a direct physical interaction .
the networks discussed above have complex topology .
spectacular advances in computerisation of data acquisition ( the human genome project is a prime example ) have made it possible to construct large databases which contain information on the topology of real life networks .
the advent of powerful computers has given rise to extensive investigations of networks containing millions of nodes .
the interesting fact emerging out of these studies is that biological networks share common topological features with non - biological networks .
there appears to be a general blueprint for the large scale organisation of several of these networks . in section 2 of this review , we discuss two types of biological networks , namely , the metabolic networks of several organisms and the protein - protein interaction networks associated with the yeast s. cerevisiae @xcite and the human gastric pathogen h. pylori @xcite . the major topological features of these networks are described and the similarity in the design principles of large - scale biological and non - biological networks pointed out .
gene regulatory networks are the most significant examples of biological networks .
gene expression and regulation are the central activities of a living cell @xcite .
genes are fragments of @xmath1 molecules and determine the structure of functional molecules like @xmath2 and proteins .
in each cell , at any instant of time , only a subset of genes present is active in directing @xmath3 / protein synthesis .
the gene expression is `` on '' in such a case .
the information present in the gene is expressed through the processes of transcription and translation . during transcription
, the sequence along one of the strands of the @xmath1 molecule is copied onto a @xmath3 molecule ( @xmath4 ) .
the sequence of the @xmath4 molecule is then translated into the sequence of amino acids , which determines the functional nature of the protein molecule produced . in a gene regulatory network ,
the protein encoded by one gene can regulate the expression of other genes .
these genes in turn produce new regulatory proteins which control still other genes .
a protein may also regulate its own level of production through an autoregulatory feedback process .
the occurrence of cell differentiation , when an organism grows from its embryonic stage , depends upon the selective switching on of gene expression in individual cells .
all these cells have identical sets of genes but follow different developmental pathways depending upon the patterns of gene expression in the cells .
thus distinct types of cells such as hair and skin cells are obtained .
gene expression is also regulated in metabolism and progression through the cell cycle as well as in responses to external signals .
infected cells can multiply because the expression of certain genes is `` on '' in these cells whereas in normal cells the expression of the same genes is `` off '' . despite a vast amount of experimental data , the complex dynamical processes involved in gene regulation
are not fully understood as yet .
a large number of theoretical studies has been undertaken @xcite but only a few of these make quantitative predictions in agreement with experimental results .
two key concepts which emerge out of the theoretical studies are : nonlinearity of the network dynamics and the role of stochasticity in gene expression and regulation @xcite .
the variables of interest in the network dynamics are the concentrations of the @xmath5 , proteins and other biomolecules within the cell .
the rate of change in the concentration of a biomolecular species is a nonlinear function of the other variables .
the dynamics is governed by a set of coupled non - linear differential equations which in most cases are solved numerically .
let @xmath6 be the concentration of , say , a particular type of protein in the cell .
the rate of change of @xmath6 is given by @xmath7 ( production - loss / decay ) of @xmath6 per unit time .
+ the production term is a nonlinear function of the other concentration variables and the loss term is usually proportional to @xmath6 . in section 3 of this review , we briefly describe the major features of nonlinearity in the dynamics of gene networks .
there is currently a significant emerging trend to utilise the concepts and techniques of nonlinear physics in the actual construction of synthetic gene regulatory networks with a variety of applications . in section 3 , a specific example of this , namely
, the genetic toggle switch @xcite will be given .
the biochemical rate equations which govern the dynamics of gene regulatory networks are deterministic in nature .
many molecules associated with the networks have low intracellular concentrations and consequently fluctuations in reaction rates are considerably large .
gene expression involves a series of biochemical reactions and due to stochastic fluctuations in the reaction rates , proteins are produced in short bursts at random time intervals . in the last few years , there is an increasing realization that stochasticity plays a significant role in biological processes @xcite . to give an example ,
consider the situation in which two independently produced regulatory proteins a and b are in competition to control a developmental switch that selects between two pathways depending on which protein wins .
the protein concentrations have to reach effective levels in order to activate the switch .
due to stochastic fluctuations , the amounts of proteins a and b produced as a function of time can vary widely from cell to cell . in some cells ,
protein a reaches the effective level first and activates the developmental switch along one specific pathway . in the other cells ,
protein b takes control and the other pathway is activated .
thus even a clonal cell population exhibits phenotypic variations as the cells follow different developmental pathways .
environmental signals can bias the probabilities of path choice in a regulatory circuit .
organisms make use of this mechanism to increase the probability of survival in a hostile environment .
cells often utilise fluctuations ( noise ) to randomize choices of developmental pathways when such randomization is desirable for the survival and growth of the organism .
a well - known example is that of the phage @xmath8 lysis - lysogeny network @xcite .
the bacterial e.coli cells , when infected by the virus phage @xmath8 , can follow two developmental pathways : lysis and lysogeny . in the lysogenic state ,
the infection is dormant .
phage @xmath8 is inert and integrated into the host cell s chromosome .
it replicates along with the bacterial dna and each new cell contains the dormant phage . in the lytic state , the infection proliferates .
the viral dna replicates using the host cell machinery giving rise to a large number of progeny phage .
these in turn lyse or burst the host bacterium cell and the infection spreads to more cells .
again , due to stochastic fluctuations , the cell population divides into two subpopulations : lysogenic and lytic .
the selection of a developmental pathway after the host cell is infected is not deterministic but probabilistic . in section 4 of this review
, the effect of stochasticity on the dynamics of the @xmath8-phage network will be briefly discussed .
the network illustrates the competitive control of a developmental switch by two regulatory proteins .
as already mentioned before , gene expression / regulation involves several biochemical reactions with appreciable stochastic fluctuations .
gillespie @xcite has proposed a monte carlo simulation algorithm to describe the kinetics of coupled stochastic reactions .
this method is physically more rigorous than the conventional differential equation approach .
the inherent assumption in the latter method is that the temporal changes in the concentrations of reacting molecules are both continuous and deterministic .
the assumption is not true if the concentrations are small and the reaction rates slow or if the system undergoes large , rapid and discrete transitions . in the gillespie algorithm
, changes in the numbers of the reacting molecules occur in integral numbers brought about by random , distinct reaction events .
the gillespie algorithm is described in detail in section 4 and some illustrative examples are given .
recent experiments at the level of a single cell have shown that gene expression occurs in abrupt stochastic bursts @xcite .
further , in an ensemble of cells , the levels of proteins produced have a bimodal distribution . in a large fraction of cells ,
the gene expression is either off or has a high value .
we have proposed a model of gene expression the essential features of which are stochasticity and cooperative binding of rna polymerase , the molecule responsible for transcription @xcite .
the model can reproduce the bimodality observed in experiments .
we include a description of the model in section 4 to give an additional example of the effect of stochasticity on gene regulation .
section 5 of the review contains concluding remarks .
the emphasis in this review is on recent studies of biological networks .
there are a number of exhaustive reviews and books on earlier work .
the three major themes that several recent studies focus on are : topology of networks , nonlinear dynamics and its consequences and the role of stochasticity in biological processes .
the present review is meant to be an introduction to these themes and to highlight the fact that interdisciplinary approaches are essential to develop an integrated understanding of biological networks .
many real life networks have a complex structure .
the mathematicians erds and rnyi @xcite were the first to propose a model of a complex network known as a random graph .
one starts with @xmath9 nodes and connects every pair of nodes with probability @xmath10 .
the graph thus has approximately @xmath11 links distributed in a random manner .
studies of real life networks , however , reveal that these can not be described as random graphs .
this distinction is possible on the basis of quantitative measurements of certain topological features which we define below .
several complex networks including the random graph are described as small world networks @xcite .
the small world idea implies that though the networks are large in size ( the number of nodes in a network is a measure of its size ) , any pair of nodes can be connected by a short path .
the distance between two nodes is given by the number of links along the shortest path connecting the nodes . in figure 1 , the distance between the nodes a and b is three .
the diameter of the network , also known as the average path length l , is the average of the distances between all pairs of nodes .
the global population is huge but still we live in a small world as any random pair of individuals are connected to each other through a short path of intermediate acquaintances .
this was first established by stanley milgram @xcite who found out that the average path length of intermediate acquaintances is six . in a small world network ,
the diameter scales as the logarithm of the number of nodes .
the second measurable topological feature of a complex network is its degree distribution @xcite .
the number of links by which a node is connected to the other nodes varies from node to node .
let @xmath12 be the probability that a randomly selected node has exactly @xmath13 links .
equivalently , @xmath12 is the fraction of nodes , on an average , which has exactly @xmath13 links .
one can define an average degree @xmath14 of the network , the degree of a node being the number of links attached to the node . in a random graph ,
the links are established randomly and most of the nodes have degrees close to @xmath14 .
the degree distribution @xmath12 vs. @xmath13 is poissonian .
it is strongly peaked at @xmath15 and has an exponential decay for large k , i.e. , @xmath16 for @xmath17 and @xmath18 . in many real life networks ,
the degree distribution @xmath12 has no well - defined peak but has a power - law distribution @xmath19 where the exponent @xmath20 is a numerical constant .
such networks are known as scale - free networks because they are not tied to a specific scale .
@xmath12 has a finite value over a wide range of @xmath13 values .
the power - law form of the degree distribution implies that the networks are extremely inhomogeneous unlike in the case of a random graph . in a scale- free network
, there are many nodes with few links and a few nodes with many links .
the highly connected nodes play a key role in the functionality of the network . both the random graph and the scale - free networks are small world networks . the third topological quantity which is measurable
is known as the clustering coefficient @xcite .
the coefficient is a measure of the tendency of the nodes of the network towards clustering . in a social network ,
the individuals are the nodes and two nodes are connected by a link if the individuals are acquainted with each other . in such a network ,
one s friend s friends are also likely to be one s friends giving rise to a clustering of acquaintances .
the clustering coefficient is defined in the following manner .
let us select a specific node @xmath21 in the network which is connected by @xmath22 links to @xmath22 other nodes . if these first neighbours are all part of a cluster , there would be @xmath23 links between them .
the clustering coefficient @xmath24 of node i is given by @xmath25 where @xmath26 is the number of actual links which exist between the @xmath22 nodes .
the clustering coefficient c of the whole network is obtained by taking an average over all the @xmath24 values .
the utility of the clustering coefficient is demonstrated in the following example .
the neural network of the nematode worm c. elegans is small in size @xcite .
the number of neurons which constitute the nodes of the network is 282 .
a link exists between two nodes if the neurons are connected by either a synapse or a gap junction .
the average degree of the network is @xmath2714 .
now consider a random graph of the same size and average degree .
the average path lengths for the neural network and the random graph are similar , 2.65 and 2.25 respectively .
is the neural network then a random graph ?
the answer is no as the clustering coefficient of the former has the value 0.28 which is much larger than the value 0.05 in the case of the latter network .
examples of real life networks which are scale - free are @xcite : the collaboration graph of movie actors ( size @xmath9 of the network = 212 250 nodes , average degree @xmath27 28.78 , the exponent @xmath20 in eq.(1 ) is @xmath20 = 2.3 ) , the www ( @xmath28 325 729 , @xmath27 5.46 , @xmath29 2.1 ) and the network of citations ( @xmath28 783 339 papers , @xmath27 8.57 , @xmath29 3 ) .
the results are obtained from available databases .
a more comprehensive and up to date list of networks is given in ref .
@xcite .
we now discuss some complex , biological networks .
recently , jeong et al @xcite have systematically investigated the topological properties of the core metabolic networks of 43 different organisms representing all the three domains of life .
the data on these organisms are available in the wit ( what is there ) database .
as already mentioned in the introduction , the nodes of the metabolic network are the different substrates .
two substrates are connected by a link if they participate in the same biochemical reaction .
the metabolic networks have different sizes , the less complex organisms having smaller sizes .
there is considerable variation in the individual constituents and the pathways of the networks .
yet they display identical topological scaling properties which resemble those of complex non - biological networks .
the metabolic networks have been found to belong to the class of scale - free networks .
the probability that a substrate participates in k reactions has a power - law distribution .
the links in a metabolic network are directed as many biochemical reactions are preferentially catalysed in one direction . for each node
, one has to distinguish between incoming and outgoing links . correspondingly , there are two exponents @xmath30 and @xmath31 .
the exponents turn out to have the same value of 2.2 . in the metabolic network ,
the distance between two substrates is given by the number of links ( reactions ) in the shortest biochemical pathway connecting the two substrates . a surprising result obtained by jeong et al
is that the diameter of the metabolic network is the same for all the 43 organisms , i.e. , it does not depend upon the number of substrates ( nodes ) belonging to the network .
this is counterintuitive and only possible if with increasing organism complexity pre - existing individual substrates are increasingly connected in order to maintain a more or less constant network diameter . in support of this conjecture ,
jeong et al found that the average number of reactions in which a certain substrate participates increases as the number of substrates in the organism increases .
conservation of the network diameter may be favourable for the survival and growth of an organism .
a larger diameter would possibly diminish the organism s ability to respond to changes in an efficient manner .
the scale - free character of the metabolic network implies that a few hubs which are highly connected play a dominant role in the functioning of the network . on sequential removal of these highly - connected nodes ,
the network diameter rises sharply and ultimately the network disintegrates into isolated fragments . on the other hand ,
the network diameter does not change appreciably when the nodes with a few links are removed from the network .
scale - free networks , in general , are robust against random mutations / errors but vulnerable to attacks targeted at highly connected nodes .
complex communication networks are surprisingly robust , local failures rarely hamper the global transmission of information .
organisms can grow and survive in hostile environments due to the error tolerance of the underlying metabolic network .
a random graph is not as robust against random mutations / errors .
mutagenesis studies in - silico and in - vivo @xcite have established the remarkable error tolerance of the metabolic network of e.coli on removing a large number of metabolic enzymes .
jeong et al , in their study of the metabolic networks of organisms found that only @xmath32 of all the substrates present in all the 43 organisms are present in all the species .
the striking fact is that the small number of substrates , common to all species , turn out to be the most highly connected ones . on the other hand , there are species - specific differences in the case of less - connected substrates .
jeong et al in a separate study @xcite have investigated the protein - protein interaction network of the yeast s.cerevisiae .
the network has 1870 proteins as nodes which are linked by 2240 direct physical interactions identified mostly by systematic two - hybrid experiments .
actual measurements show that the probability @xmath12 that a given yeast protein interacts with @xmath13 other yeast proteins has a power - law distribution with an exponential cutoff at @xmath33 20 .
@xmath34 with @xmath35 1 and @xmath29 2.4 .
the protein - protein interaction network of the bacterium h. pylori @xcite displays similar topology .
for the metabolic networks , the exponent @xmath20 has the value 2.2 .
the value of @xmath20 falls in the range 2.0 - 2.5 for many scale - free networks . like the metabolic network and other scale - free networks , the protein - protein interaction network is found to be immune to random mutations .
the removal of highly connected nodes may , however , disrupt the network function .
the protein product of the p53 tumor - suppressor gene is one of the most highly connected proteins found in human cells .
mutations of p53 gene affect cellular functions severely from a biomedical point of view .
in fact , jeong et al s study on the protein - protein interaction network in yeast shows that proteins with five or lesser number of links constitute @xmath36 93% of the total number of proteins but only @xmath36 21% of them are essential so that the removal of such proteins proves to be lethal . in contrast , only @xmath36 0.7% of the total number of proteins have more than 15 links but single deletion of @xmath36 62% of these severely affects the functioning of the network .
it is possible that the proteins which constitute the highly connected nodes in a network share common structural features .
these features favour the binding of many different types of proteins to the proteins in question .
the scale - free character of both the metabolic and protein - protein interaction networks suggests the evolutionary selection of a common large scale structure of biological networks .
studies of other biological networks are expected to provide further evidence for this idea .
the dynamics of gene regulatory networks are described by coupled nonlinear p.d.e.s which can be collectively represented as @xmath37 where @xmath38 is the @xmath9 -component state vector @xmath39 and @xmath40 is a set of nonlinear functions @xmath41 .
there are thus n coupled p.d.e.s and an individual p.d.e .
is of the form @xmath42 there are in total n species of biochemical molecules participating in m reactions .
@xmath44 represents the concentration of the ith molecular species at time @xmath45 .
r represents a set of control parameters .
the functions @xmath46 are nonlinear functions of the @xmath47 and the specific forms of the functions are determined by the structures and rate constants of the m chemical reactions . as an example consider the set of reactions @xmath48 @xmath49
@xmath50 @xmath51 the reactions represent the conversion of the precursor species @xmath52 into a final product @xmath53 via a sequence of four reactions involving two intermediates @xmath54 and @xmath55 .
the third reaction is autocatalytic as @xmath55 catalyses its own production .
the second reaction represents the uncatalysed conversion of @xmath54 to @xmath55 and the last reaction shows that the catalyst @xmath55 decays into the product @xmath53 .
the equations are assumed to be irreversible .
also , the concentration of the reactant @xmath52 is assumed to be constant over a reasonable period of time .
this is possible if the initial concentration of @xmath52 is large .
let @xmath56 ( constant ) , @xmath57 and @xmath58 denote the concentrations of the molecular species @xmath52 , @xmath54 and @xmath55 . the decay product @xmath53 does not participate in any further reaction and so does not influence the chemical kinetics .
the rates of the four successive chemical reactions are @xmath59 , @xmath60 , @xmath61 and @xmath62 respectively where @xmath63 , @xmath64 , @xmath65 , @xmath66 are the rate constants .
the equations governing the chemical kinetics are @xmath67 @xmath68 in the general scheme of p.d.e.s shown in eq .
( 5 ) , @xmath9 = 2 , i.e. , there are two molecular species @xmath54 and @xmath55 participating in @xmath69 = 4 chemical reactions . @xmath70 and @xmath71 are the concentrations of the moleculeas @xmath54 and @xmath55 . the r.h.s.s of eqs .
( 10 ) and ( 11 ) are the nonlinear functions @xmath72 and @xmath73 the nonlinearity arising from the autocatalytic term @xmath61 .
the rate constants together with @xmath56 constitute the control parameters @xmath74 .
in the general case , imagine an abstract @xmath9 - dimensional state space with axes @xmath75 .
the state of the system at any instant of time , say @xmath76 , is given by the @xmath9 -component state vector @xmath77 . in the state space , this state is represented by a single point .
the time evolution of the system gives rise to a trajectory in the state space .
the trajectory may end up at a fixed point @xmath78 . at this point , the rates of change of all the variables in the system are exactly zero , i.e. , the l.h.s.s of the @xmath9 equations in eq.(5 ) are zero .
the system is said to be in the steady state at the fixed point . at this point ,
the state of the system remains unchanged as a function of time .
the only way of changing the state of the system is to apply perturbations to it .
a fixed point is stable if small perturbations around the point eventually damp out .
the stable fixed point acts as an attractor to the states in its vicinity .
the corresponding region in the state space is called the basin of attraction .
the nonlinear dynamics may give rise to more than one fixed point .
if there are two stable fixed points , the system is bistable , i.e. , two stable steady states are possible .
one can similarly define multistability .
the other long - term possibilities for the trajectory in the state space are a limit cycle and a strange attractor . in the first case
, the trajectory goes towards a closed loop and eventually circulates around it forever . in physical terms
, this corresponds to stable oscillations in the system .
the strange attractor is a set of states to which the trajectory is confined , never stopping or repeating .
such aperiodic motion is often indicative of chaos in the system .
we now discuss the role of the control parameters r ( eqs . ( 4 ) and ( 5 ) ) in the nonlinear dynamics of a system . by varying these parameters
, one can bring about changes in the qualitative structure of the dynamics .
such changes are known as bifurcations .
for example , as a parameter is changed , a steady state can become unstable and replaced by stable oscillations . a system with one stable steady state changes over to multistability , i.e.
, the system can exist in multiple steady states . to give a simple example of bifurcation , consider the rate equation @xmath79 there are two fixed points of this equation : @xmath80 = 0 and @xmath81 . to determine the stability of the fixed points
, one undertakes what is known as the linear stability analysis .
one determines the time evolution of a small perturbation @xmath82 around the fixed point . by substituting @xmath83 in eq.(12 ) and ignoring terms of the order of @xmath84 , one obtains @xmath85 when @xmath80 = 0 .
the fixed point is stable if @xmath86 since @xmath87 reduces to zero during time evolution .
the fixed point is unstable if @xmath88 > 0 and @xmath89 = 0 is the bifurcation point . if @xmath81 , then @xmath90 .
hence the fixed point is unstable if @xmath88 < 0 and stable for @xmath88 > 0 .
different types of bifurcation are possible a detailed discussion of which is given in standard textbooks and reviews @xcite on nonlinear dynamics .
if there is more than one stable fixed point , a switch - like behaviour is possible . in the case of bistability ,
the system remains in one stable state until a sufficiently large perturbation drives the system to the other stable state .
the system continues to remain in the latter state even after the perturbation is removed .
the @xmath8-phage lysis - lysogeny network offers an example of bistability @xcite .
the lytic and the lysogenic states are the two possible steady states . a transition from the lysogenic to the lytic state occurs on irradiating with ultra - violet light . in a gene regulatory network , a negative ( positive )
feedback implies that a gene product inhibits ( promotes ) its own level of activity . to give an example , a protein which represses the transcription of its own gene operates through negative feedback .
it has been found that negative ( positive ) feedback increases ( decreases ) stability in gene regulatory systems @xcite .
real life gene regulatory networks are often complex .
some of the examples are the @xmath8-phage lysis - lysogeny circuit , the regulatory network for the activation of the tumour - suppressor protein p53 @xcite and the bacteriophage t7 ( another lytic phage which infects e.coli ) network @xcite .
computational modelling studies of these networks have been undertaken with a view to explain experimental results .
the quantitative agreement between theory and experiment is most often not good .
the reasons are two fold : the complex nature of the networks and the difficulty in carrying out actual experiments on them
. computational as well as mathematical modelling of simpler networks is more extensive .
such networks incorporate the essential features of their more complex counterparts .
the models seek to explain experimental results at a qualitative level .
there are also abstract mathematical models of gene expression / regulation which highlight the general principles and their outcomes .
there are already some good reviews and books on the computational and mathematical modelling of gene regulatory networks @xcite . for the purpose of this review ,
we pick on just one example , that of a synthetic gene regulatory network which illustrates the importance of nonlinearity in the dynamics of the network .
gardner et al @xcite have constructed and tested a synthetic , bistable gene regulatory network based on the predictions of a simple mathematical model .
the network is called a genetic toggle switch and consists of two repressors ( proteins ) and two promoters . the enzyme rna polymerase ( rnap )
binds to the promoter region of a dna sequence to initiate the process of transcription .
the initial binding of rnap to a promoter can be prevented by the binding of a regulatory protein to an overlapping segment of dna , called operator .
the gene expression is off in this case .
2 shows a simple sketch of the toggle network .
the two promoters are designated as @xmath91 and @xmath92 .
@xmath91 drives the expression of the @xmath93 gene and @xmath92 that of the @xmath94 gene .
the @xmath93 and @xmath94 genes express the proteins of the same names .
the proteins mutually inhibit the production of each other , hence the name repressor .
the @xmath93 proteins form tetramers and the tetramer binds to operator sites adjacent to the @xmath92 promoter , blocking the transcription of the @xmath94 gene in the process .
the @xmath94 proteins , when produced , form dimers .
the repressor dimer cooperatively binds to the operator sites in the vicinity of the @xmath91 promoter . as a result ,
transcription of the @xmath93 gene is not possible .
the nonlinear dynamics of the toggle network are governed by the following two equations : @xmath95 @xmath96 where @xmath6 and @xmath97 are the concentrations of @xmath93 and @xmath94 proteins respectively , @xmath98 and @xmath99 are the effective rates of synthesis of @xmath93 and @xmath94 proteins , @xmath100 is the cooperativity of repression of the @xmath91 promoter and @xmath20 the same in the case of the @xmath92 promoter .
3 reveals the origin of bistability in the system .
the nullclines @xmath101 = 0 and @xmath102 = 0 intersect at three points .
these are the fixed points ( steady states ) of the dynamics .
two of the fixed points are stable and the third unstable .
the bistability occurs provided @xmath103 > 1 ( cooperative repression of transcription ) and the rates of synthesis of the two repressors are balanced .
if the rates are not balanced , the nullclines intersect at a single point giving rise to a single stable steady state ( monostability ) . in the region of bistability
, the two stable steady states correspond to ( 1 ) state 1 ( high @xmath97 / low @xmath6 ) and ( 2 ) state 2 ( low @xmath97 / high @xmath6 ) respectively .
there are two basins of attraction , one above the separatrix and the other below it . in the @xmath104 vs. @xmath105 parameter space ,
bifurcation lines separate the monostable and bistable regions @xcite .
the size of the bistable region decreases on reducing the cooperativity of repression ( @xmath100 and @xmath20 ) .
the parameters @xmath106 and @xmath20 act as the control parameter r changing which a transition ( bifurcation ) between monostability and bistability occurs . in the region of bistability
, the toggle is flipped between the stable states ( states 1 and 2 ) using transient chemical or thermal induction .
the chemical agent isopropyl-@xmath100 -d - thiogalactopyranoside ( iptg ) can bind to @xmath93 tetramers . as a result ,
the latter can not bind to the operator region in the neighbourhood of the promoter @xmath92 , i.e. , @xmath93 can no longer repress the production of the @xmath94 proteins .
suppose the bistable system is originally in the stable state 2 ( high @xmath6 ( @xmath93 ) , low @xmath97 ( @xmath94 ) ) .
on the induction of iptg , the concentration of the @xmath94 proteins increases as a function of time .
the @xmath94 proteins in their turn repress the production of @xmath93 proteins the concentration of which begins to fall .
the dynamics ultimately leads the system to the other fixed point ( state 1 ) .
the system remains in this stable steady state ( low @xmath6 / high @xmath97 ) even after the removal of the iptg stimulus .
how can the toggle flip back to state 2 ?
this is achieved by using a temperature - sensitive @xmath94 protein in the network .
the degradation rate of this protein increases as temperature is raised . on raising the temperature to @xmath107 ( actual experiment ) ,
the concentration of @xmath94 proteins starts to fall .
since repression is less , the concentration of @xmath93 proteins starts to go up .
the system finally reaches the fixed point corresponding to the stable steady state 2 . after the steady state
is reached , the temperature of the system is reduced ( @xmath108 in the experiment ) . the system continues to remain in the steady state 2 .
a full cycle of the switching process is now completed .
the actual construction of the toggle switch has been accomplished in e.coli using the standard tools of molecular biology @xcite .
there is a reasonable agreement between the theoretical predictions based on eqs.(13 ) and ( 14 ) and the results obtained from experiments on the synthetic toggle network .
the design of the network relies significantly on theoretical inputs like identification of the region of bistability , increasing the cooperativity in repression ( @xmath100 and @xmath20 ) to achieve bistability over a wider region in parameter space etc . as a practical device
, the toggle switch may have applications in biotechnology , biocomputing and gene therapy . as a cellular memory unit
, the toggle provides the basis for `` genetic applets '' which are self - contained , programmable synthetic gene networks used in the control of cell functions . in parallel with the toggle work , another synthetic network , the repressilator has been designed and tested @xcite .
the repressilator dynamics is again nonlinear and give rise to oscillations in the concentrations of the cellular proteins .
the design of the network is based on a simple mathematical model of transcriptional regulation .
the repressilator provides insight about the design principles of other oscillatory systems such as circadian clocks found in many organisms including cyanobacteria . the genetic toggle switch and the repressilator
demonstrate that theoretical models can provide the design criteria for the actual construction of synthetic , gene regulatory networks .
these simple networks have applications as practical devices and also help us to understand the functional properties of the more complex , naturally - occurring networks .
nonlinear dynamics can give rise to various types of instability one of which is the turing instability . in 1952 ,
turing @xcite in a seminal paper proposed a mechanism for pattern formation in biological systems as well as the development of structure during the growth of an organism .
examples of biological patterns are the spots on the skin of a leopard , the stripes of a zebra , the arrangement of veins on the leaves of a tree etc . structure formation is initiated by the process of cell differentiation , an example is the emergence of limbs in an organism during the growth of the organism from the featureless embryonic stage .
the turing mechanism involves both reaction as well as diffusion processes . to illustrate the mechanism , consider two chemical agents , the activator and the inhibitor .
the activator is autocatalytic , i.e. , it promotes its own production as well as that of the inhibitor .
the inhibitor , as the name implies , is antagonistic to the activator and represses its production .
both the chemicals can diffuse but the inhibitor has a much larger diffusion coefficient . consider a homogeneous distribution of the activator and the inhibitor in the system . increase the concentration of the activator by a small amount in a local region .
this gives rise to further increases in the local concentrations of the activator and the inhibitor .
the inhibitor quickly diffuses to the surrounding region and prevents the activator from reaching there .
thus , in the steady state , islands of high activator concentration exist in a sea of high inhibitor concentration .
the islands constitute what is known as the turing pattern .
diffusion in general smooths out concentration differences in a system but the turing process involving both reaction and diffusion gives rise to a steady pattern of concentration gradients .
there is now increasing evidence that chemical gradients play a crucial role in the formation of patterns and cell differentiation in biological systems . to give an example
, the protein bicoid has been found to have a graded concentration distribution in the drosophila melanogaster embryo .
it is responsible for the organization of the anterior half of the fly and has been fully characterised @xcite .
many reaction - diffusion ( rd ) models have been proposed based on the turing mechanism and some of these can reproduce the patterns observed in nature @xcite .
the basic scale of a pattern is larger than the size of an individual cell and so the rd processes involve more than one cell .
cells possibly choose developmental pathways depending upon their location in the concentration gradient .
position - dependent activation of genetic swtches in the cells may constitute an important step in both pattern and structure formation .
direct evidence for this in terms of a detailed characterization of the genes involved and an identification of the actual biochemical reactions occurring in the cells , is , however , yet to be obtained .
turing patterns have so far been experimentally observed in certain chemical rd systems in the laboratory @xcite and also in some biological systems @xcite .
as already mentioned in the introduction , stochastic fluctuations in the dynamics of the gene regulatory network lead to a probabilistic selection of developmental pathways .
the @xmath8-phage lysis - lysogeny network @xcite was discussed as an example .
4 shows some of the key components of the network .
the complexity of the full network is captured in figure 1 of ref .
we confine our attention to the simpler network .
it consists of two @xmath8 -phage genes @xmath94 and @xmath109 .
the corresponding promoters are @xmath110 and @xmath111 respectively .
transcription of the gene @xmath94 ( @xmath109 ) expresses the regulatory protein @xmath8 repressor ( @xmath112 ) .
both the proteins are capable of binding to the operator regions @xmath113 , @xmath114 and @xmath115 .
they act antagonistically to control promoter activity .
transcription of the @xmath94 gene , initiated from the promoter @xmath110 , takes place whenever there is no protein of either type binding to @xmath115 .
the @xmath8 repressor molecule has a dumbell shape and there is a tendency for two such molecules to bind and form a dimer .
the operator region @xmath113 has the highest affinity for the binding of @xmath8 repressor dimer .
the binding increases the affinity of @xmath114 for a second repressor dimer , i.e. , a cooperative binding of dimers to the operator regions @xmath113 and @xmath114 takes place .
the @xmath8 repressor has both negative and positive control . if the @xmath8 repressor is present at @xmath114 , transcription of the @xmath109 gene is not possible .
this is because the repressor covers part of the @xmath1 that a @xmath116 molecule must have access to in order to recognize the promoter @xmath111 , bind to it and initiate the transcription of the @xmath109 gene .
the same repressor at @xmath114 exhibits positive control in helping a rnap molecule to bind to the promoter @xmath110 and begin transcription of the @xmath94 gene .
the increase in the transcription rate is approximately tenfold @xcite .
if @xmath114 is not occupied by the repressor , the transcription rate of the @xmath94 gene is low .
the reason for the dramatic increase in the transcription rate is the following .
the presence of a repressor dimer bound to @xmath114 leads to an increased affinity of @xmath110 for @xmath116 because the polymerase is held at @xmath110 not only by its contacts with the @xmath1 but also due to the protein - protein contact with the repressor . in summary ,
a repressor dimer bound to @xmath114 , represses transcription from @xmath111 but promotes transcription at @xmath110 .
the @xmath109 gene is transcribed only when the operator region @xmath115 is either empty or has @xmath112 dimer bound to it .
the transcription of the @xmath94 gene can not take place if the @xmath113 and @xmath114 regions are occupied by either protein , @xmath8 repressor and @xmath112 . in the lysogenic state ,
all the phage genes are off except for one gene @xmath94 which produces the protein @xmath8 repressor .
the protein in turn binds to the operators @xmath113 and @xmath114 in the form of dimers and activates the transcription of its own gene at @xmath110 .
the bound @xmath8 repressor dimers further prevent transcription initiation at @xmath111 .
irradiation of the lysogen with ultra - violet light inactivates @xmath8 repressor making the synthesis of the second regulatory protein @xmath112 possible .
@xmath112 promotes lytic growth and competes with the @xmath8 repressor in occupying the same operator sites .
increased @xmath112 production leads to a greater probability of @xmath112 binding at @xmath115 which prevents the initiation of transcription at @xmath110 .
the concentration of the @xmath8 repressor starts to fall down as a result .
the concentration of @xmath112 proteins increases and when it reaches a level such that the operator regions @xmath113 and @xmath114 begin to be occupied , the transcription at @xmath111 is also halted .
the switchover from the lysogenic to the lytic state is further possible by @xmath117 -mediated degradation of the @xmath8 repressor ( @xmath117 is a catalytic protein ) .
arkin et al @xcite have analysed the stochastic kinetics of the full @xmath8-phage network which consists of more genes and regulatory elements than shown in figure 4 .
their detailed investigations show that fluctuations in the rates of gene expression give rise to random patterns of protein production in individual cells and wide diversity in instantaneous protein concentrations across cell populations .
each cell has two developmental pathways : lytic and lysogenic .
the pathway selection depends upon which protein , @xmath8 repressor or @xmath112 , takes control of the operator region .
if it is the @xmath8 repressor , the lysogenic pathway is chosen .
if the @xmath112 takes control , the lytic pathway is selected . due to stochastic fluctuations , the concentrations of @xmath8 repressor and @xmath112 vary considerably from cell to cell tipping the balance in favour of one or the other pathway . as a result
, initially homogeneous cell populations can partition randomly into distinct lytic and lysogenic subpopulations .
arkin et al have constructed a stochastic kinetic model of the @xmath8-phage circuit and based on model calculations predicted the fraction of infected cells selecting the lysogenic pathway at different phage : cell ratios .
the theoretical results are consistent with the experimental results of kourilsky @xcite .
the kinetic model uses the stochastic formulation of chemical kinetics @xcite , stochastic mechanisms of gene expression @xcite and a statistical - thermodynamical model of promoter regulation @xcite .
probabilistic selection of developmental pathways occurs in several other gene regulatory networks producing stochastic phenotypic outcomes .
some examples are given in table 4 of ref .
@xcite .
we now describe the well - known gillespie algorithm @xcite which is incresingly being used by biologists in the stochastic kinetic approach to the study of gene expression and regulation in different systems let us consider a system of @xmath9 chemicals participating in @xmath69 reactions @xmath118 .
the state of the system at any instant of time @xmath45 is represented as @xmath119 where @xmath120 is the number of molecules of the ith chemical species .
two questions have to be answered to determine how the system evolves in time : ( 1 ) when will the next reaction occur and ( 2 ) what type of reaction will it be ?
let @xmath121 = the probability that an @xmath118 @xmath122 reaction occurs in the next infinitesimal time interval @xmath123 for a particular combination of the reactant molecules .
let @xmath124 be the number of distinct combinations of molecules available in the state @xmath125 for the @xmath118 reaction .
+ as an example , consider the reaction @xmath126 let @xmath127 and @xmath128 be the number of molecules of types a and b respectively .
then @xmath129 . let @xmath130 @xmath131 be the probability that an @xmath118 reaction occurs in time @xmath132 given the system is in the state @xmath119 at time @xmath45 .
+ the reaction probability density function @xmath133 is the probability that given the state @xmath119 at time @xmath45 , the next reaction will occur in the infinitesimal time interval @xmath134 and will be an @xmath118 reaction , @xmath135 where @xmath136 is the probability that no reaction occurs in the time interval @xmath137 and @xmath138 is the subsequent probability that an @xmath118 reaction occurs in the time interval @xmath134 .
now @xmath139\ ] ] where the expression inside the bracket is the probability that no reaction occurs in time @xmath140 from the state @xmath141 .
( 17 ) can be solved to obtain @xmath142\ ] ] substituting for @xmath136 in eq .
( 16 ) , one gets @xmath143 if @xmath144 , @xmath145 and @xmath146 otherwise . @xmath147 and @xmath148 now the goal is to generate a pair of random numbers @xmath149 acording to the probability distribution ( 19 ) .
to do this , use the standard random number generator to obtain two random numbers from the uniform distribution in the unit interval .
take @xmath150 and @xmath88 is chosen to be the integer for which @xmath151 the pair of numbers @xmath149 , ( eqs .
( 22 ) and ( 23 ) ) , belongs to the set of random pairs described by the probability density function @xmath152 . for a rigorous proof of this see refs .
once @xmath149 are known , put @xmath153 and adjust the @xmath120 values according to the @xmath118 reaction .
if the @xmath118 reaction is the one shown in eq.(15 ) , both @xmath127 and @xmath128 have to be decreased by 1 and @xmath154 the number of molecules of c , increased by 1 .
the input values at time @xmath45 = 0 are @xmath155 and the initial values of @xmath156 .
the steps of the gillespie algorithm are : step 1 + calculate @xmath157 and @xmath158 .
step 2 + generate @xmath159 and @xmath160 with the help of a uniform random number generator .
calculate @xmath161 and @xmath88 according to the formulae in eqs .
( 22 ) and ( 23 ) .
step 3 + advance @xmath45 by @xmath161 ( eq.(24 ) ) and adjust the @xmath120 values according to @xmath118 .
then repeat the steps from step 1 to further advance the system in time .
recently , kierzek et @xcite al have used the gillespie algorithm to study a stochastic kinetic model of prokaryotic gene expression .
they explicitly considered ten biochemical reactions : @xmath162 @xmath163 @xmath164 @xmath165 where @xmath52 denotes the promoter region of the gene and @xmath166 the bound promoter-@xmath116 complex .
reaction 2 ( eq.(26 ) ) describes @xmath116 dissociation and reaction 3 the isomerization of `` closed complex '' to `` open complex '' , @xmath167 is the activated @xmath168promoter complex .
reaction 4 describes clearance of promoter region by @xmath169 @xmath170 stands for @xmath116 transcribing the gene and synthesizing the @xmath4 molecule and @xmath171 is the ribosome binding site on @xmath4 .
the other reactions are : @xmath172 @xmath173 @xmath174 @xmath175 @xmath176 @xmath177 reactions 5 - 10 ( eqs .
( 29)-(34 ) ) describe translation , @xmath4 decay and protein degradation .
reaction 5 describes @xmath178 binding to @xmath171 , the bound complex is designated as @xmath179 .
reaction 6 is the dissociation of the bound complex .
reaction 7 describes @xmath178 binding site clearance , @xmath180 is the @xmath178 which translates the @xmath4 .
reaction 8 has degradation of @xmath171 by the enzyme @xmath181 .
@xmath181 and @xmath182 are in competition to occupy @xmath171 .
if @xmath181 binds first then it initiates the degradation of @xmath4 but does not interfere with the movement of the already bound @xmath182 engaged in the process of translation .
every @xmath178 which successfully binds to the @xmath171 completes translation of the protein .
reaction 9 corresponds to the completion of protein synthesis .
reaction 10 represents protein decay .
the stochastic rate constants @xmath183 of the different reactions , needed as inputs to the gillespie algorithm , can be calculated from the more familiar chemical rate constants listed in kierzek et al s paper @xcite . for first order chemical reactions , the stochastic rate constant is equal to the rate constant of a chemical reaction . for second order reactions , the stochastic rate constant is equal to the rate constant divided by the volume of the system ( in this case a cell ) . in the last part of this section , we describe a cooperative stochastic model of gene expression proposed by us @xcite . as already explained in the introduction , the model has been constructed to explain the bimodal distribution in gene expression observed in recent experiments .
the model describes the transcription of a single gene with one promoter region .
there is one operaor region to which a regulatory protein @xmath74 can bind .
this prevents the binding of a @xmath116 to the promoter so that transcription of the gene can not be initiated .
there is a finite probability that the bound @xmath74 molecule dissociates from the operator at any instant of time .
@xmath116 molecule then has a certain probability of binding to the promoter and initiating transcription .
each of the possibilities described above actually involves a series of physico - chemical processes , a detailed characterization of which is not required for the model of gene expression proposed by us .
we represent a gene by a one - dimensional lattice of @xmath184 sites .
the first two sites represent the operator and promoter respectively .
the lattice is a coarse - grained description of an actual gene . in reality
, the operator and promoter regions may extend over a certain number of base pairs in the dna and they can be overlapping or not . in our model , they are represented as single sites .
each of the other sites in the lattice represents a finite number of base pairs in the dna molecule .
the different physico - chemical processes are lumped together into a few simple events which are random in nature .
this lumping together avoids unnecessary complexity that has no bearing on the basic nature of the process .
the operator @xmath185 and the promoter @xmath186 together can be in four possible configurations : @xmath187 and @xmath188 .
the numbers `` @xmath189 '' and `` @xmath190 '' stand for `` occupied '' and `` unoccupied '' .
the configuration @xmath191 describes the occupation status of @xmath192 @xmath193 and @xmath52 @xmath194 .
for example , the configuration @xmath195 corresponds to @xmath192 being occupied by a @xmath74 molecule and @xmath52 being unoccupied .
similarly , in the configuration @xmath196 , @xmath192 is unoccupied and @xmath52 is occupied by a @xmath116 molecule .
binding of @xmath74 and @xmath116 molecules are mutually exclusive so that the configuration @xmath188 is strictly prohibited . given a @xmath197 configuration at time @xmath45 , the transition probabilities to configurations @xmath195 and @xmath196 at time @xmath198 are @xmath199 and @xmath200 respectively .
the probability of remaining in the configuration @xmath197 is @xmath201 .
a @xmath195 configuration at time @xmath45 goes to a @xmath197 configuration at time @xmath198 with probability @xmath202 and remains unchanged with probability @xmath203 .
we have assumed all the probabilities to be time - independent .
the @xmath116 molecule once bound to the promoter initiates transcription in the next time step , i.e. , the @xmath196 configuration makes a transition to a @xmath197 configuration with probability @xmath189 .
the motion of @xmath116 is in the forward direction and the molecule covers a unit distance ( the distance between two successive lattice sites ) in each time step .
once the molecule reaches the last site of the lattice , the transcription ends and a @xmath4 is synthesized .
the second major feature of our model is the cooperative binding of @xmath116 to the promoter , when an adjacent @xmath116 molecule is present .
this implies that there is a higher probability of binding of @xmath116 to the promoter in one time step if another @xmath116 molecule is present at the site next to the promoter . in our model ,
the probability of cooperative binding is @xmath204 which is larger than @xmath200 .
the probabilities @xmath199 and @xmath201 are changed to new values @xmath205 and
@xmath206 respectively . degradation of @xmath4 is taken into account by assuming the decay rate to be given by @xmath207 , where @xmath9 is the number of @xmath5 present at time @xmath45 .
the number of @xmath5 produced as a function of time is studied by monte carlo simulation .
for the sake of simplicity , we have not tried to simulate protein levels or enzymatic products thereof , i.e. , we study gene expression upto the level of transcription ( @xmath4 synthesis ) .
since the number of protein molecules and converted products should be proportional to the number of @xmath4 molecules , no loss of generality is introduced by this simplification .
the lattice consists of 52 sites @xmath208 .
stochastic events are simulated with the help of a random number generator .
the updating rule of our cellular automaton ( ca ) model is that in each time step @xmath45 , the occupation status ( @xmath190 or @xmath189 ) of each site ( except for the @xmath192 site ) at time @xmath209 is transferred to the nearest - neighbour site towards the right .
if the last site is @xmath189 at @xmath209 , a @xmath4 is synthesized at @xmath45 and the number of @xmath5 increases by one . in the same time step ,
the configuration @xmath191 of @xmath210 is determined with the probabilities already specified .
thus , in each time step , the @xmath116 molecule , if present on the gene , moves forward by unit lattice distance ( progression of transcription ) followed by the updating of the @xmath210 configuration .
figure 5 shows the concentration @xmath211 $ ] of @xmath4 molecules in the cell as a function of time for the parameter values @xmath212 and @xmath213 .
note that an almost four - fold increase in the probability of @xmath116 binding is assumed due to cooperativity .
the stochastic nature of the gene expression is evident from the figure , with random intervals between the bursts of activity .
one also notices the presence of several bursts of large size .
it is important to emphasize that the frequency of transition between high and low expression levels is a function of the parameter values chosen and may be low for certain parameter values . for the probability values considered ,
the two predominantly favourable states are when the gene expression is off ( state 1 ) and when a large amount of gene expression takes place ( state 2 ) . in the absence of @xmath116 binding , state 1 has greater weight but with the chance binding of @xmath116 to the promoter ( probability @xmath200 for this is small ) , the weight shifts to state 2 until another stochastic event terminates cooperative binding and the gene reverts to state 1 .
the probability of obtaining a train of @xmath9 successive transcribing @xmath116 molecules is @xmath214 .
this is the geometric distribution function and the mean and the variance of the distribution are given by @xmath215 and @xmath216 respectively . for the probability values
already specified , the simulation has been repeated for an ensemble of 3000 cells .
for each cell , the time evolution is upto 10000 time steps .
figure 6 shows the distribution of the number @xmath217 of cells versus the fraction @xmath218 of the maximal number of @xmath4 molecules produced after @xmath219 time steps .
two distinct peaks are seen corresponding to zero and maximal gene expression .
such a bimodal distribution occurs over a range of parameter values .
some theories have been proposed so far to explain the so - called `` all or none '' phenomenon in gene expression @xcite .
these theories are mostly based on an autocatalytic feedback mechanism , synthesis of the gene product gives rise to the transport or production of an activator molecule . while such processes are certainly possible , the bimodal distribution is a much more general phenomenon and has now been found in many types of cells , from bacterial to eukaryotic and for different types of promoters @xcite .
the two major features of the model of gene expression that we have proposed are stochasticity and cooperative binding of @xmath116 .
there is by now enough experimental evidence of stochasticity in gene expression .
our suggestion of cooperative binding of @xmath116 is novel and there is no direct experimental verification of the proposal as yet .
there are some recent experiments which provide indirect evidence and these are discussed in ref .
in this review we have given an elementary introduction to some of the major aspects of biological networks , namely , topological characteristics , nonlinear dynamics and the role of stochasticity in gene expression and regulation .
the main aim of the review is to highlight the usefulness of interdisciplinary approaches in the study of both natural and synthetic biological networks
. some important features of such networks have not been discussed in the review .
one of these is the operational reliability of networks in spite of randomness in basic regulatory mechanisms .
many regulatory pathways do have highly predictable outcomes even when stochastic fluctuations are considerable .
cells adopt various strategies like populational transcriptional cooperation , checkpoints to ensure that cascaded events are appropriately synchronised , redundancy and feedback to achieve regulatory determinism .
some of these ideas are discussed in ref .
the complexity of biological networks raises the question of their functional stability .
in particular , the issue of interest is the sensitivity of networks to variations in their biochemical parameters .
barkai and leibler @xcite have studied a biochemical network responsible for bacterial chemotaxis and shown that the functional properties of the network are robust , i.e. , relatively insensitive to changes in biochemical parameters like reaction rate constants and enzymatic concentrations .
bialek @xcite has shown that extremely stable biochemical switches can be constructed from small numbers of molecules though intuitively one expects such systems to be prone to instability due to the inherent noise .
metzler @xcite in a recent paper has shown that spatial fluctuations in the distribution of regulatory molecules play a non - trivial role in genetic switching processes .
apart from internal stochastic fluctuations , external noise originating in random variations in the environment or in the externally set control parameters , may affect the functioning of a biological network .
hasty et al @xcite have proposed a synthetic genetic network in which external noise is utilised to operate a protein switch ( short noise pulses are used to turn protein production `` on '' and `` off '' ) . in another novel application , external noise is used to amplify gene expression , i.e. , protein production by a considerable amount .
genetic networks with many components are difficult to analyze using conventional techniques .
many parallels have been drawn in the functioning of genetic and electrical circuits @xcite . in electrical engineering , there are well developed techniques of circuit analysis which can be used to characterise the operation of complex electrical networks . some of these techniques are increasingly being used to study genetic networks .
engineers are familiar with some of the design principles of biological networks . rapid transitions between the two stable states of a system can be brought about by positive feedback loops .
negative feedback loops control the value of an output parameter to be within a narrow range even if there are wide fluctuations in the input .
coincidence detection systems activate an output provided two or more events occur simultaneously .
parallel connections enable a device to remain functional in the event of failures in one of the lines .
one can give analogous examples from biology .
one set of positive feedback loops is responsible for the rapid transition of cells into mitosis ( division of cell nucleus ) , another set brings about the exit from mitosis in an irreversible manner .
gene transcription in eukaryotes involve coincidence detection .
a @xmath4 can be produced only if the promoters regulating gene expression are occupied by the different transcription factors .
these examples indicate that general principles govern the functioning of genetic and electrical networks though there are other aspects of such networks which are not common to both .
biological networks constitute a field of research the interdisciplinary nature of which will become more evident as we progress into the twenty first century . + * acknowledgement : the author thanks subhasis banerjee for help in drawing the figures . * s. h. strogatz , nature 410 , 268 ( 2001 ) r. albert and a .-
barabsi , cond - mat/0106096 , to appear in rev .
s. n. dorogovtsev and j. f. f. mendes , cond - mat/0106144 , to appear in adv .
h. jeong , b. tombor , r. albert , z.n .
oltvai and a .-
barabsi , nature 407 , 651 ( 2000 ) h. jeong , s.p .
mason , a .-
barabsi and z.n .
oltvai , nature 411 , 41 ( 2001 ) p.
uetz et al , nature 403 , 623 ( 2000 ) j .- c .
rain et al , nature 409 , 211 ( 2001 ) b. levin , genes v ( oxford university press , new york 1994 ) j. hasty , d. mcmillen , f. isaacs and j.j .
collins , nature reviews genetics 2 , 268 ( 2001 ) h. h. mcadams and a. arkin , trends in genetics 15 , 65 ( 1999 ) t. s. gardner , c.r .
cantor and j. j. collins , nature 403 , 339 ( 2000 ) h. mcadams and a. arkin , proc .
94 , 814 ( 1997 ) a. arkin , j. ross and h. h. mcadams , genetics 149 , 1633 ( 1998 ) d. t. gillespie , j. comput .
22 , 403 ( 1976 ) d. t. gillespie , j. phys .
81 , 2240 ( 1977 ) g. zlokarnik et al , science 279 , 84 ( 1998 ) p.a .
negulescu , n. shastri and m.d .
cahalan , proc .
91 , 2873 ( 1994 ) j. karttunen and n. shastri , proc .
88 , 3972 ( 1991 ) a. novick and m. weiner , proc .
43 , 553 ( 1957 ) s. roy , i. bose and s. s. manna , int . j. mod . phys .
c 12 , 413 ( 2001 ) p. erds and a. rnyi , publ .
inst . hung .
5 , 17 ( 1960 ) m. e. j. newman , j. stat .
101 , 819 ( 2000 ) s. milgram , psychology today 2 , 60 ( 1967 ) d. j. watts and s. h. strogatz , nature 393 , 440 ( 1998 ) a .- l . barabsi and r. albert ,
science 286 , 509 ( 1999 ) j. s. edwards and b. o. palsson , proc .
97 , 5528 ( 2000 ) s. h. strogatz , nonlinear dynamics and chaos ( perseus , new york 1994 ) s. wiggins , introduction to applied nonlinear dynamical systems and chaos ( springer , new york 1990 ) m. c. cross and p. c. hohenberg , rev .
65 , 851 ( 1993 ) m. a. savageau , nature 252 , 546 ( 1974 ) b. vogelstein , d. lane and a. j. levine , nature 408 , 307 ( 2000 ) d. endy , l. you , j. yin and i. j. molineux , proc .
97 , 5375 ( 2000 ) g. rowe , theoretical models in biology ( clarendon press , oxford 1994 ) ; j. j. tyson and h. g. othmer , prog .
theor . biol . 5 , 1 ( 1978 ) u. s. bhalla and r. iyengar , science 283 , 381 ( 1999 ) m. b. elowitz and s. leibler , nature 403 , 335 ( 2000 ) ;
see also j. hasty , f. isaacs , m. dolnik , d. mcmillen and j. j. collins , chaos 11 , 207 ( 2001 ) a. m. turing , philos .
london , ser .
b 237 , 37 ( 1952 ) w. driever and ch .
nsslein - volhard , cell 54 , 83 ( 1988 ) l. boring , m. weir and g. schubiger , mech .
42 , 97 ( 1993 ) a. j. koch and h. meinhardt , rev .
66 , 1481 ( 1994 ) h. meinhardt , models of biological pattern formation ( academic , new york 1982 ) j. dziarmaga , physics/0002050 i. bose and i. chaudhuri , phys .
e 55 , 5291 ( 1997 ) ; i. bose and i. chaudhuri , int .
c 12 , 247 ( 2001 ) v. castets et al , phys .
64 , 2953 ( 1990 ) q. ouyang and h. l. swinney , nature 352 , 610 ( 1991 ) s. kondo and r. asai , nature 376 , 765 ( 1995 ) s. sawai , y. maeda and y. sawada , phys .
85 , 2212 ( 2000 ) m. ptashne , a genetic switch : phage @xmath8 and higher organisms ( cell press , cambridge , massachusetts 1992 ) p. kourilsky , mol .
122 , 183 ( 1973 ) m.a .
shea and g.k .
ackers , j. mol .
181 , 211 ( 1985 ) a.m. kierzek , j. zaim and p. zielenkiewicz , j. biol .
276 , 8165 ( 2001 ) m. t. beckman and k. kirkegaard , j. biol .
273 , 6724 ( 1998 ) t. a. carrier and j.d .
keasling , j. theor .
201 , 25 ( 1999 ) n. barkai and s. leibler , nature 387 , 913 ( 1997 ) w. bialek , cond - mat/0005235 r. metzler , phys .
87 , 68103 ( 2001 ) j. hasty , j. pradines , m. dolnik and j. j. collins , proc .
97 , 2075 ( 2000 ) h. h. mcadams and l. shapiro , science 269 , 650 ( 1995 ) l. h. hartwell , j. j. hopfield , s. leibler and a. w. murray , nature 402 ( supplement ) , c47 ( 1999 ) + figure captions + figure 1 .
example of a network .
the solid circles and lines denote the nodes and the links respectively . + figure 2 .
schematic diagram of the synthetic genetic toggle switch network .
there are two promoters @xmath91 and @xmath92 and two genes @xmath93 and @xmath94 .
+ figure 3 .
graphical representation of the toggle equations ( eqs .
( 13 ) and ( 14 ) ) .
@xmath220 are the concentrations of the @xmath93 and @xmath94 proteins respectively .
there are two stable steady states : state 1 ( high @xmath97/ low @xmath6 ) and state 2 ( low @xmath97/ high @xmath6 ) and one unstable steady state .
+ figure 4 .
some key components of the @xmath8-phage lysis - lysogeny network : @xmath221 are the two genes , @xmath110 and @xmath111 are the two promoters and @xmath222 and @xmath115 are the three operator regions .
+ figure 5 .
concentration of @xmath4 molecules @xmath211 $ ] in arbitrary units as a function of time t. the parameter values are @xmath212 and @xmath213 .
+ figure 6 .
distribution of no .
@xmath217 of cells expressing fraction @xmath218 of maximal number of @xmath4 molecules produced after 10000 time steps .
the total number of cells is 3000 .
the parameter values are as in figure 5 .
|
in this review , we give an introduction to the structural and functional properties of the biological networks .
we focus on three major themes : topology of complex biological networks like the metabolic and protein - protein interaction networks , nonlinear dynamics in gene regulatory networks and in particular the design of synthetic genetic networks using the concepts and techniques of nonlinear physics and lastly the effect of stochasticity on the dynamics .
the examples chosen illustrate the usefulness of interdisciplinary approaches in the study of biological networks .
department of physics , bose institute , 93/1 , a.p.c .
road , calcutta-700009 , india
|
the timeline of publications for different journals and overall is presented in fig .
it can be observed that the overall output ( bottom most color stripe ) increases steadily over time .
an obvious exception is the world war ii period , during which the production dropped almost an order of magnitude , from nearly 100 publications per month before and after the war to below 10 during the war .
this confirms , not surprisingly , that periods of war decelerate scientific progress or at least very much hinder the dissemination of new knowledge . by geocoding the affiliations ,
it is also possible to infer where the physics published in the physical review has been coming from . as can be inferred from fig .
[ geo ] and the accompanying video , the formative years of physical review were dominated by the us , with relatively rare contributions coming from the uk , germany , france and india . during the world wars i and ii large contingents of the world went silent ( see the video referred to in the caption of fig .
[ geo ] ) , and it was only during the 1950s and 60s that the us centrism begun fading .
the collapse of the soviet union , the fall of the berlin wall , and the related changes in world order during the 1980s and 90s contributed significantly to the globalization , so that today countries like china , russia , canada , japan , australia , as well as large regions of europe and south america all contribute markedly to physics research that is published in the physical review . countries that are still exempt are from central africa and sahara .
ranking the countries according to their overall average monthly production during the last 20 years yields us , germany , france , uk , japan , italy , china , russia , spain , canada and netherlands , while per capita yields switzerland , israel , denmark , sweden , slovenia , finland , germany , netherlands , france and austria as the top 10 , respectively .
these results are in good agreement with more comprehensive rankings that were recently published based on world citation and collaboration networks over many different fields of research @xcite .
our goal here is solely to provide a general overview of the geographical origin of the data , and so we proceed with the core analysis of trends of physics discovery .
to do so , we employ the methodology described in the methods section .
results for the word `` quantum '' are presented in fig .
[ time ] ( the n - gram viewer for publications of the american physical society is available at http://www.matjazperc.com/aps [ ] ) , where the vertical lines denote the starting times of windows during which the maximal upward and downward trends were recorded . by performing the same analysis on the whole data set and ranking the trends in a decreasing manner ( we use absolute values for negative slopes @xmath0 ) , we arrive at the biggest ever upward and downward movers across the whole publishing history of the american physical society . since
the obtained tables are too big to be displayed meaningfully in print , we make them available online at http://www.matjazperc.com/aps/rankings [ ] , separately for all time windows @xmath1 and eligible journals .
the physical review st : physics education research and physical review x ( prx ) do not have an extensive enough publication history to qualify for this analysis .
although it would be interesting to comment on the trends of particular words and phrases and reconcile them with other historical accounts , the options for how to do that are simply too many to be meaningfully covered in this publication .
we hope readers will find their favorites amongst the trendsetters and conduct experiments of their own . here
we proceed with the focus on the large - scale properties of the trends . as fig .
[ dist ] demonstrates , the distributions of magnitude have heavy tails , largely independent of the direction of trend and journal .
nevertheless , subtle differences can be inferred , and they deserve special attention .
to determine the properties of the depicted distributions more accurately , we test several hypotheses .
the first is that the depicted cumulative distributions follow a power law @xmath2 . by using maximum - likelihood fitting methods and goodness - of - fit tests based on the kolmogorov - smirnov statistics @xcite
, we find that only for the journals depicted bold in the legend of fig .
[ dist ] the power law is an acceptable fit .
the distributions of downward trends for the physical review ( pr ) and physical review series i ( pri ) are best described by a power law with an exponential cut - off @xmath3 , while the distribution of upward trends for the physical review e ( pre ) is a stretched exponential @xmath4 .
the pertaining exponents are summarized in table [ fit ] .
.[fit ] characterization of the distribution of the magnitude of trends reveals that for the majority the power law is an acceptable hypothesis .
journals for which both the upward and downward trends follow a power law are denoted bold .
we use @xmath5 as the threshold for acceptance @xcite .
the distributions of downward trends for pr and pri are described best by a power law with an exponential cut - off having ( @xmath6 , @xmath7 ) and ( @xmath8 , @xmath9 ) , respectively , while the distribution of upward trends for pre is a stretched exponential with ( @xmath10 , @xmath11 ) . for the distribution of upward trends for the reviews of modern physics ( rmp ) and for both distributions concerning physical review st : accelerators and beams ( prstab ) , none of the five considered functions , including the exponential and log - normal in addition to the three already mentioned , provide an acceptable fit . [ cols="^,^,^",options="header " , ] although distributions depicted in fig .
[ dist ] are not the most beautiful power laws , and some altogether fail to conform to the power - law hypothesis , the prevalence of heavy tails nevertheless hints firmly towards robust large - scale self - organization governing the up and down trends . by defining the trend rate as @xmath12 , where @xmath13 is the smallest time interval between two consecutive trajectory points
, we can directly test for the matthew effect .
however , since the trajectories exhibit both up and down trends , we determine the upward and downward rates separately within time windows of maximal growth and decline .
results presented in fig . [ rate ] confirm that the more commonly used a given word or a phrase is , the larger its expected upward momentum is going to be .
the same holds true for the magnitude of falls during times of decline .
together with the continuity of scientific progress , the matthew effect gives rise to strong but rare trendsetters on the expense of the majority of discoveries that remain forever unknown except to those that made it .
the model of growth and preferential attachment @xcite captures the essence of our observations .
attachment rates that are linearly proportional with the degree of each node translate into power - law distributions , while deviations from the linear form lead to deviations in the corresponding distributions .
near - linear attachment rates yield log - normal distributions @xcite , while sublinear attachment rates yield distributions with an exponential cut - off or stretched exponential distributions @xcite , depending further on the details of sublinearity .
the accuracy of empirical studies will also be impaired by finite - size effects and saturation , which may additionally contribute to deviations from the power law @xcite . by contrasting the distributions in fig .
[ dist ] with the corresponding rates depicted in fig
. [ rate ] , we find an agreement that is well aligned with the theoretical expectations . moreover , having a closer look at the journals for which the deviations from the linear rates are particularly strong , we find either that they were published in a time when abstracts were rare ( pri , partly also pr ) , that their publication history is relatively short ( pre , prstab ) , or that they publish reviews rather than original research ( rmp ) , all of which are probable causes for the analysis on this particular cases to give less conclusive results .
the identified self - organization in the rise and fall of scientific paradigms can be seen akin to previous reports of preferential attachment in citation rates and the acquisition of scientific collaborators @xcite .
specifically related to the former case , our discovery can be interpreted as the textual extension of the matthew effect in citation rates or as the large - scale `` semantic '' version of that effect .
it is also worth noting that , although it is debatable whether the concept of preferential attachment is based on luck or reason @xcite , in our case at least it seems inevitably due to the actual progress made , not chance that could make one discovery seem bigger than it truly is .
after identifying all unique words and phrases , we determine their relative frequency of occurrence @xmath14 with respect to the number of publications in any given month for each journal published by the american physical society as well as overall .
we consider a phrase to be a string of words separated by a space , and we limit our analysis to at most four - word phrases to keep the volume of information to be processed manageable . by ignoring capitalization , numbers , words containing numbers , and formulae , we identify 118056 single words , 3269090 two - word phrases , 13295156 three - word phrases , and 23799449 four - word phrases , thus obtaining over 40 million trajectories that enable a qualitative exploration of the trends of physics discovery . while of course not all identified words and phrases have to do with physics , the assumption we make is that only those that do will actually exhibit notable trends .
words like `` the '' or `` of '' appear in nearly every abstract .
the word `` quantum '' , on the other hand , is mentioned first in the 1917 may issue of the physical review , with popularity subsequently peaking in january 1927 at @xmath15 , as depicted in fig . [ time ] .
trajectories of all other words and phrases can be searched for and viewed at http://www.matjazperc.com/aps [ ] . however , since not all publications of the american physical society have an abstract , and since some abstracts are very short , even the most common words and phrases can occasionally exhibit relatively strong trends . not to treat those trends as trends of physics discovery
, we eliminate from the analysis the most common english words as identified in @xcite , minus a few hand - picked special cases that obviously have to do with physics .
we also eliminate phrases that contain the most common english words either at the beginning or end . with these two additional filters in place , we make sure that from all the identified unique words and phrases the focus is on those that , in the majority of cases , concern at least some aspect of physics . to quantify the trends , we seek out time windows where the slope @xmath0 of the linear fit of each trajectory is maximally positive and maximally negative , and we do so separately for windows of width @xmath16 , @xmath17 , @xmath18 and @xmath19 years .
the dispersal in years is important as we want our analysis to encompass short- , mid- and long - term trends .
although a straight line wo nt be a good fit for the data in several cases , it is nevertheless a useful first - order approximation for whether a subject is trending up or down , and to what extent this is the case @xcite .
recent most advances on how to identify trends in word frequency dynamics are presented in @xcite , and they shall be an excellent basis for future explorations .
as starting points of each of the four considered time windows , we consider every month of every word and phrase for which data is available , with the obvious condition that the starting point plus the window width must not go beyond october 2012 . before the analysis we apply a 12 month moving average on the trajectories and
require that the considered time windows must not contain missing data after averaging .
moreover , we dismiss all words and phrases with @xmath20/year as lacking notable trends .
csermely , p. , korcsmaros , t. , kiss , h. j. m. , london , g. , and nussinov , r. structure and dynamics of molecular networks : a novel paradigm of drug discovery : a comprehensive review .
preis , t. , kenett , d. y. , stanley , h. e. , helbing , d. , and ben - jacob , e. quantifying the behavior of stock correlations under market stress .
, 752 ( 2012 ) .
|
we make use of information provided in the titles and abstracts of over half a million publications that were published by the american physical society during the past 119 years . by identifying all unique words and phrases and determining their monthly usage patterns , we obtain quantifiable insights into the trends of physics discovery from the end of the 19th century to today . we show that the magnitudes of upward and downward trends yield heavy - tailed distributions , and that their emergence is due to the matthew effect .
this indicates that both the rise and fall of scientific paradigms is driven by robust principles of self - organization .
data also confirm that periods of war decelerate scientific progress , and that the later is very much subject to globalization . ] the 20th century is often referred to as the century of physics @xcite . from x - rays to the semiconductor industry ,
the human society today would be very different were it not for the progress made in physics laboratories around the world . and while amid the economic woes the budget for science is being cut down relentlessly @xcite , it seems now more than ever the need is there to remind the policy makers of this fact .
although to the layman the progress made on an individual level may appear to be puny and even needless , the history teaches us that collectively the physics definitively delivers .
it is therefore of interest to understand how the progress made so far came to be , and how to best maintain it in the future .
should there be overarching authorities that dictate which scientific challenges to address and prioritize , or should we rely on the spontaneous emergence of progress ? we know
, for example , that the acquisition of citations @xcite as well as the acquisition of collaborators @xcite are subject to preferential attachment .
these two processes are neither regulated nor imposed .
they are perpetuated by scientific excellence and individual choice .
in fact , barabsi and albert @xcite have shown that preferential attachment and growth give rise to robust principles of self - organization that culminate in the emergence of scaling .
preferential attachment can be considered synonymous to the matthew effect , which sociologist robert k. merton @xcite coined based on the writings in the gospel of st .
matthew for explaining discrepancies in recognition received by eminent scientists and unknown researchers for similar work .
derek j. de solla price @xcite observed the same phenomenon when studying the network of citations between scientific papers , and most recently also the longevity of careers has been found driven by the matthew effect @xcite .
motivated by the existing reports of the matthew effect in science , we explore whether the trends of scientific discovery are also subject to the same principles of self - organization .
we make use of the titles and abstracts of over half a million publications of the physical review that were published between july 1893 and october 2012 , and we infer the trends by adopting the methodology of culturomics @xcite .
our approach is thus purely data - driven @xcite , in line with substantial interdisciplinary research efforts that are currently aimed at obtaining quantitative insights into the social and natural sciences in general @xcite , but also into sports @xcite , drug discovery @xcite , finance @xcite , and scientific production @xcite in particular .
|
the study of superfluidity and superconductivity in two - dimensional ( 2d ) fermi systems has attracted great attention in the past several decades , partly because of its close relationship to the problem of high-@xmath0 superconductors @xcite .
recently , the experimental progress on creating quasi - low - dimensional atomic gases in optical lattices @xcite and on atom chips @xcite provides us a possibility of realizing and studying superfluidity in a more controllable platform . in particular , with the aid of tuning an external magnetic field through a feshbach resonance , the interaction between fermionic atoms can be tuned continuously from the bardeen - cooper - schrieffer ( bcs ) limit to the bose - einstein condensation ( bec ) limit , so that the bcs - bec crossover can be studied @xcite .
the bcs - bec crossover in a 2d fermi system has been discussed at both zero @xcite and finite temperatures @xcite , where an effective 2d hamiltonian with atom - atom interaction ( called model 1 ) is employed . however , for strongly interacting fermionic atoms in a realistic quasi-2d geometry , due to the inevitable population of many excited levels along the strongly confined transverse direction @xcite , one needs to introduce a composite particle called dressed molecule to account for the population in these transverse excited levels , and write down a more complicated form for the effective 2d hamiltonian ( called model 2 ) with renormalized interaction between atoms ( in the transverse ground level ) and dressed molecules @xcite .
we have shown that the model 1 and the model 2 hamiltonians lead to qualitatively distinct results in some cases @xcite .
for instance , for quasi-2d fermions in a weak global harmonic trap , the model 1 hamiltonian predicts a constant value of the thomas - fermi cloud size at zero temperature from the bcs to the bec limit .
in contrast , the model 2 hamiltonian predicts that the cloud size shrinks as one approaches the bec limit @xcite , which is the correct trend as one should expect from the bcs - bec crossover picture . in this paper , we extend the discussion of the model 2 hamiltonian to finite temperatures by taking into account the fluctuation effects .
in particular , we consider a trapped quasi-2d fermi gas and study the superfluid transition temperature across a wide range of feshbach resonance .
note that according to the mermin - wagner - hohenberg - coleman theorem @xcite , phase fluctuations are dramatic at any finite temperatures such that there will be no ordinary long - range order in two dimensions . in this case , the superfluid transition is not accompanied by a bose condensation , but instead characterized by a berezinskii - kosterlitz - thouless ( bkt ) type , for which the vortex - antivortex pairs start to form at the transition and a topological order is established @xcite . using the model 2 hamiltonian
, we show that in a uniform quasi-2d fermi gas , the bkt transition temperature @xmath1 increases from zero at the bcs limit and approaches monotonically to almost a constant value about @xmath2 at the bec limit , where @xmath3 is the 2d fermi energy .
furthermore , we take into account the weak harmonic trap in the radial plane , and investigate the number and superfluid density distributions under the local density approximation ( lda ) . as a characteristic feature of a bkt transition ,
a finite jump of superfluid density is present in the middle of the trap , which takes a universal value of @xmath4 @xcite .
a quasi-2d fermi gas is typically prepared in experiments by means of a one - dimensional ( 1d ) deep optical lattice along the axial ( @xmath5 ) direction , where tunneling between different lattice sites is completely suppressed .
a 1d lattice is described by the potential @xmath6\cos ( k_{z}z)$ ] , where @xmath7 is the wave vector of the laser beam , and @xmath8 is the waist width satisfying @xmath9 .
this optical lattice potential can be approximated around the minimal points by a strongly anisotropic pancake - shaped harmonic potential with @xmath10 , where @xmath11 , @xmath12 , and @xmath13 is the radial distance .
the strong anisotropy of the trap ( @xmath14 ) introduces two well separated energy scales , which allows us to first deal with the axial degrees of freedom by deriving an effective 2d hamiltonian , while the radial degrees of freedom are left for later treatment with the lda approximation .
the effective 2d hamiltonian describing a quasi-2d fermi gas within a strongly axial confinement is derived in ref . @xcite . under natural units ( @xmath15 ) ,
the hamiltonian takes the form @xmath16 where @xmath17 and @xmath18 are the annihilation operators for fermionic atoms and bosonic dressed molecules , respectively .
the dressed molecule consists of a tight pair of atoms distributed in the transverse ( axial ) excited levels as well as the population in the closed feshbach channel @xcite . in eq .
( 1 ) , @xmath19 is the dispersion relation for fermions with mass @xmath20 and radial momentum @xmath21 , @xmath22 is the fermionic chemical potential , @xmath23 labels the pseudo - spin , and @xmath24 is the quantization area . in the remainder of this manuscript ,
we choose @xmath25 as the energy unit so that @xmath26 , @xmath22 , and @xmath27 become dimensionless , where @xmath28 is the characteristic length scale for axial motion .
the 2d effective bare parameters @xmath29 ( detuning ) , @xmath30 ( atom - molecule coupling rate ) , and @xmath31 ( background interaction ) in eq .
[ eq : tc - hamiltonian ] are connected with the 2d physical parameters @xmath32 , @xmath33 , and @xmath34 through the renormalization relation @xmath35 ^{-1 } & \equiv & \left [ v_{p}+\frac{% \alpha _ { p}^{2}}{x-\lambda _ { p}}\right ] ^{-1 } \notag \\ & = & \left [ v_{b}+\frac{\alpha _ { b}^{2}}{x-\lambda _ { b}}\right ] ^{-1}+\frac{1}{% l^{2}}\sum_{\mathbf{k}}\frac{1}{2\epsilon _ { \mathbf{k}}+\omega _ { z}},\end{aligned}\ ] ] which is an identity for the variable @xmath36 . the inverse of the physical effective interaction @xmath37^{-1}$ ]
are determined from the three - dimensional ( 3d ) atomic scattering data through @xcite @xmath38 ^{-1}=\frac{\sqrt{2\pi } } { % a_{z}^{2}}\left [ \left ( u_{p}+\frac{g_{p}^{2}}{x-\gamma _ { p}}\right ) ^{-1}-s_{p}(x)+\sigma _ { p}(x)\right ] , \label{eq : vbneg}\ ] ] where @xmath39 , @xmath40 , and @xmath41 are 3d dimensionless parameters . here , @xmath42 is the background scattering length , @xmath43 is the difference in magnetic moments between the open and closed collision channels , @xmath44 is the resonance width , and @xmath45 is the resonance point . the functions in eq .
( [ eq : vbneg ] ) are kestner-07 @xmath46 , \label{eq : sp } \\
\sigma _ { p}(x ) & \equiv & \frac{\ln \left\vert x\right\vert } { 4\pi \sqrt{2\pi } % } , \label{eq : sigma}\end{aligned}\ ] ] where @xmath47 is the gamma function . the partition function @xmath48 of the hamiltonian eq . (
[ eq : tc - hamiltonian ] ) at temperature @xmath49 can be expressed as an imaginary - time functional integral with action @xmath50 , \label{eq : action1}\ ] ] where @xmath51 and @xmath52 become imaginary time dependent variables . by introducing the hubbard - stratonovich field @xmath53 , which couples to @xmath54 , and integrating out the fermions
, we obtain @xmath55\exp ( -s_{% \mathrm{eff}})$ ] , with the effective action given by @xmath56-\mathbf{tr}\left [ \ln \mathbf{g}_{\mathbf{k},\mathbf{k}% ^{\prime } } ^{-1}\right ] .\end{aligned}\ ] ] notice that the trace @xmath57 is taken over momentum , imaginary time , and nambu indices of the inverse nambu matrix @xmath58 , with @xmath59here , @xmath60 is the field whose stationary value corresponds to the order parameter .
we expand the action around a stationary and homogeneous saddle point @xmath61 .
the saddle point equation is derived from the saddle point action @xmath62 $ ] via the stationary conditions @xmath63 .
this process leads to @xmath64 ^{-1}=-\frac{1}{l^{2}}\sum_{% \mathbf{k}}\left [ \frac{1}{2e_{\mathbf{k}}}\tanh \left ( \frac{\beta e_{% \mathbf{k}}}{2}\right ) -\frac{1}{2\epsilon _ { \mathbf{k}}+1}\right ] , \label{eq : gapft}\ ] ] where @xmath65 is the quasi - particle excitation spectrum .
we have used the 2d renormalization relation eq .
( [ eqn : renorm ] ) in the derivation , which introduces the term @xmath66 . in order to study the fluctuation around the saddle points of @xmath67 , @xmath68 , and @xmath69
, we next write the field operators as [ eqn : orderparameters ] @xmath70 which are all assumed to be slowly varying in both spatial and temporal coordinates . here , @xmath71 , @xmath72 , and @xmath73 are magnitude fluctuations , while @xmath74 , @xmath75 , and @xmath76 are phase fluctuations .
the magnitude fluctuations correspond to gapped excitations , whose contribution can be neglected . however , the phase fluctuations correspond to gapless excitations , and it is their presence that breaks the ordinary long - range order in two dimensions .
notice that since the order parameter @xmath77 is a linear combination of @xmath78 and @xmath79 , neglecting its magnitude fluctuation @xmath80 is equivalent to assuming a same phase factor for @xmath18 and @xmath79 .
we thus have @xmath81 the phase fluctuation @xmath82 around the saddle points is in general not a small value , so the phase factor in the order parameter can not be directly expanded , but must be treated as a whole .
this can be done by introducing a unitary transformation @xmath83\mathcal{o}$ ] for the fields @xmath84 , @xmath79 , and @xmath77 . as a result of this transformation , and after a fourier transformation back to the coordinate space , we can expand the effective action around its stationary point of @xmath69 , leading to @xmath85+s_{\mathrm{fluc}}[\nabla \theta , \partial _
{ \tau } \theta ] , \label{eq : action3}\ ] ] where the fluctuation contribution @xmath86 contains terms only involving spatial or temporal gradients of the field @xmath87 , which are assumed to be small .
expanding @xmath86 up to the quadratic order of @xmath88 and @xmath89 leads to @xmath90 , \label{eq : action4}\ ] ] where @xmath91 represents the phase stiffness ( or called the superfluid density ) and takes the form @xmath92
\notag \label{eq : rhos } \\ & & \hspace{1cm}-\frac{\beta \hbar ^{2}}{4ml^{2}}\sum_{\mathbf{k}}k^{2}\mathrm{% sech}^{2}\left ( \frac{\beta e_{\mathbf{k}}}{2}\right ) \bigg\}.\end{aligned}\ ] ] here , @xmath93^{-1}/\partial \mu $ ] denotes the dressed - molecular population .
other parameters in eq .
( [ eq : action4 ] ) are @xmath94 .\end{aligned}\ ] ] ( solid lines ) of a uniform quasi-2d fermi gas for various values of @xmath95 .
notice that @xmath1 increases monotonically from the bcs ( right ) to the bec ( left ) side of the resonance , and approaches to a limiting value of @xmath96 . in the bec limit , @xmath97 can also be estimated by considering a weakly interacting gas of composite bosons with 3d scattering length of @xmath98 , leading to a logarithmically varying result ( dashed line ) .
the mean - field transition temperature @xmath99 ( dashed line in the inset ) is also shown as a comparison for @xmath100.,width=321 ] by decomposing the phase fluctuation @xmath101 into a vortex part @xmath102 and a vortex - free spin - wave part @xmath103 , one can separate the effective action @xmath104 , and hence the thermodynamic potential as @xcite @xmath105 where @xmath106 , @xmath107 and @xmath108 are the corresponding contributions from the saddle point action @xmath109 , the vortex contribution from @xmath102 , and the spin - wave contribution from @xmath110 , respectively .
the spin - wave part can be integrated out to give @xmath111 , where @xmath112 is the speed of the spin - wave .
the number equation hence can be obtained via @xmath113 , leading to @xmath114\end{aligned}\ ] ] with the spin - wave contribution @xmath115 . by writing the number equation as above , we implicitly assume a vanishingly small vortex density , such that the vortex contribution to the number equation can be safely neglected .
this condition is in general valid for temperatures around or below the superfluid transition temperature @xcite .
( solid lines ) as a function of @xmath116 for various values of @xmath117 .
notice that @xmath118 can exceed @xmath3 ( dotted ) even on the high - field side of the resonance point , hence pushes the system into the bec limit as indicated by the transition temperature shown in fig .
[ fig : homogeneous].,width=321 ] the @xmath91 is the superfluid density at the microscopic scale , and the vortex fluctuation tends to renormalize ( decrease ) the superfluid density when the physical scale @xmath119 goes up .
when the superfluid density is renomalized to zero with @xmath120 , the berezinskii - kosterlitz - thouless ( bkt ) transition takes phase where the system changes from a superfluid phase to a normal phase .
the renormalization - group ( rg ) flow depends on the vortex core energy @xmath121 , which can be estimated by @xmath122 @xcite .
the rg equations for the superfluid density @xmath123 and the fugacity @xmath124 are given by @xcite @xmath125 y(l)+o(y^{2 } ) , \label{eqn : rg2}\end{aligned}\ ] ] where @xmath126 .
the initial conditions are @xmath127 and @xmath128 .
the fixed point @xmath129 of these rg equations corresponds to the critical condition for the bkt transition , where the vortex- antivortex pairs start to dissociate and break superfluidity .
thus , the equation for the critical temperature @xmath130 is determined as @xcite @xmath131 where @xmath132 is the renormalized superfluid density .
this equation must be solved together with the saddle - point equation ( [ eq : gapft ] ) , the number equation ( eq : number1 ) , and the rg equations ( [ eqn : rg1 ] ) and ( [ eqn : rg2 ] ) to determine @xmath22 , @xmath69 , and @xmath1 as functions of the magnetic field detuning . as a comparison , the mean - field transition temperature @xmath99 can be calculated by solving only eqs .
( eq : gapft ) and ( [ eq : number1 ] ) with @xmath133 .
next , we will consider some specific systems and discuss the corresponding numerical results .
we first ignore the trapping potential in the radial plane , and consider a uniform quasi-2d fermi gas .
a typical set of results of @xmath1 for various values of @xmath134 are shown in fig .
[ fig : homogeneous ] . here
, we consider the case of @xmath135li and use the parameters @xmath136 , @xmath137 g , and @xmath138 , where @xmath139 and @xmath140 are bohr radius and bohr magneton , respectively .
we also notice that when plotted as functions of the inverse 3d scattering length @xmath141 , the results for @xmath142k are very close to those for @xmath135li , indicating the universal behavior around the resonance point .
notice that the superfluid transition temperature @xmath1 increases continuously from the bcs to the bec side of the feshbach resonance , and saturates to a limiting value of @xmath143 .
this limiting value of @xmath1 is expected for @xmath144 , which corresponds to the bec limit of a quasi-2d fermi gas where paired fermions behave like weakly interacting bosons with number density @xmath145 and mass @xmath146 . in the bec limit
, we can also consider the particles as composite bosons with effective three - dimensional scattering length of @xmath147 @xcite . for weakly interacting bosons ,
more accurate results are available for the bkt transition temperature based on a combination of the quantum monte carlo simulation @xcite and the renormalization group approach @xcite , which predicts that @xmath1 goes down logarithmically with the scattering length .
this more accurate result in the bec limit is also shown in fig .
[ fig : homogeneous ] , which is slightly deviant from our fermionic calculation .
we note also from fig .
[ fig : homogeneous ] that the bec limit value of @xmath148 can even be reached on the bcs side of the feshbach resonance ( with @xmath149 ) . in fact , since a two - body bound state is always present in quasi - low dimensions at arbitrary detunings , the bec limit of a quasi-2d fermi gas can be realized as long as the condition @xmath150 is satisfied , regardless of the sign of the 3d scattering length @xmath151 .
this property is in clear contrast to the 3d case , where a bound state is only present with positive @xmath151 , hence the bec limit can only be realized on the low - field side of a feshbach resonance . here ,
the two - body binding energy @xmath118 is determined by solving the hamiltonian ( [ eq : tc - hamiltonian ] ) for two particles , which gives @xmath152 ^{-1}=\frac{1}{4\pi a_{z}^{2}}\ln ( -e_{b } ) .
\label{eq : eb}\ ] ] the corresponding results for @xmath118 are plotted in fig .
[ fig : binde ] . notice that for the case with @xmath153 , the binding energy @xmath154 for @xmath155 , leading to a bkt transition temperature of the bec limit value @xmath143 , as shown in fig .
[ fig : homogeneous ] .
another feature of fig . [ fig : homogeneous ] is that the transition temperature @xmath1 increases with the axial trapping frequency @xmath156 at a fixed detuning .
this trend can also be easily understood by analyzing the two - body binding energy @xmath118 .
in fact , at a given detuning , @xmath118 increases with @xmath25 as shown in fig .
fig : binde , hence pushes the quasi-2d system further to the bec limit .
( solid lines ) plotted as functions of the two - body binding energy @xmath118 .
the results are compared with the outcome of an effective 2d hamiltonian with renormalized atom - atom interaction ( model 1 , dotted line ) , which is discussed in refs .
, width=321 ] the results of @xmath1 should also be compared with the outcome of an effective 2d hamiltonian with renormalized atom - atom interaction ( model 1 ) , which is discussed in refs .
@xcite . in fig .
[ fig : homogeneous - eb ] , we show @xmath1 calculated using both models as functions of the binding energy @xmath118 . one prediction of the model 1 is that the many - body physics ( such as the bkt transition temperature ) takes a universal behavior , which only depends on the two - body binding energy @xmath118 in the unit of @xmath3 @xcite . as shown in fig .
[ fig : homogeneous - eb ] , this is clearly not the case for the results from the model 2 , where the transition temperature @xmath1 also depends on the other energy scale such as the transverse trapping frequency @xmath25 .
notice that both of the models predict roughly the same limiting value of @xmath1 for large values of @xmath118 ( in the bec limit ) , since in that limit the system behaves like weakly interacting bosons , and @xmath1 gets very insensitive to the interaction between these composite bosons ( logarithmic dependence as mentioned above ) . , ( b ) @xmath157 , and ( c ) @xmath158 .
for each temperature , three typical detunings are considered with @xmath159 ( bcs , left ) , @xmath160 ( unitarity , middle ) , and @xmath161 ( bec , right ) , respectively .
notice the finite jump of superfluid density in the middle of the trap at intermediate temperatures .
the plots are made for @xmath162li , while the results for @xmath142k are similar .
parameters used in this plot are @xmath163 khz , @xmath164 hz , @xmath165 , and @xmath166 is the zero - temperature thomas - fermi cloud size of a quasi-2d ideal fermi gas with particle number @xmath167 . , width=340 ] after discussing the case of a uniform quasi-2d fermi gas , we next consider the trapping potential in the radial plane @xmath168 . under the local density approximation ( lda ) , we consider a position dependent chemical potential @xmath169 . here , @xmath170 is the chemical potential at the trap center , which must be determined by fixing the total particle number in the trap @xmath171 . using this technique ,
we calculate the in - trap number and superfluid density distributions in the radial plane , which are shown in fig .
[ fig : finitetintrap ] . from the left to the right , results are sequentially presented for the bcs side of the resonance ( @xmath172 ) , unitarity ( @xmath173 ) , and the bec side of the resonance ( @xmath174 ) . for each case
, temperature is varied within two orders of magnitude , showing that superfluidity is absent at high temperatures and starts to build from the trap center as decreasing @xmath175 . at low enough temperatures ,
superfluidity nearly extends to the whole trap , and the number density profile approaches the zero temperature results @xcite . furthermore , notice that there is a finite jump of superfluid density present in the middle of the trap at intermediate temperatures , as shown in panels of fig .
[ fig : finitetintrap]b .
this discontinuity in @xmath176 signatures the position where the superfluid transition takes place , and takes a universal value of @xmath177 characterizing a phase transition in the bkt universality class @xcite .
in summary , we have studied the bkt type of superfluid transition of a quasi-2d fermi gas based on an effective 2d hamiltonian with renormalized interaction between atoms and dressed molecules .
the finite temperature effect has been taken into account by incorporating phase fluctuations over the saddle point solution . using the effective hamiltonian that we derived before with the explicit parameters @xcite
, we establish the bkttransition temperature as a function of the 3d atomic scattering length , making it possible to directly compare the result with the experimental measurements .
we also compare the predictions from this effective hamiltonian ( model 2 ) with the results from model 1 ( where the 2d hamiltonian is described by atom - atom interaction with an effective scattering length @xcite ) , and they differ significantly in certain parameter regions . in particular , the universal behavior predicted by the mode l ( many - body physics depends only on the two - body binding energy ) is not the case for the model 2 , where the bkt transition temperature also has explicit dependence on the transverse trapping frequency @xmath25 .
this difference can be quantitatively tested by future experiments . under the local density approximation
, we have also calculated the in - trap number and superfluid density distributions at finite temperatures , which can be compared with experimental results in a harmonic trap .
a. j. leggett , in _ modern trends in the theory of condensed matter _ , edited by by a. pekalski and r. przystawa , ( springer - verlag , berlin , 1980 ) ; p. nozieres and s. schmitt - rink , j. low temp .
* 59 * , 195 ( 1985 ) ; c. a. r. s de melo , m. randeria , and j. r. engelbrecht , phys .
71 * , 3202 ( 1993 ) .
|
we study the superfluid transition in a quasi - two - dimensional fermi gas with a magnetic field tuning through a feshbach resonance . using an effective two - dimensional hamiltonian with renormalized interaction between atoms and dressed molecules ,
we investigate the berezinskii - kosterlitz - thouless transition temperature by studying the phase fluctuation effect .
we also take into account the trapping potential in the radial plane , and discuss the number and superfluid density distributions .
these results can be compared to experimental outcomes for gases prepared in one - dimensional optical lattices .
|
let us first summarize the present values of neutrino masses and mixing angles obtained by fitting atmospheric , solar , reactor and accelerator neutrino oscillation data .
the atmospheric and solar neutrino masses and mixing angles are given below * @xmath0 @xmath1 * @xmath2 @xmath3 with an approximate mixing matrix given by @xmath4 for this talk , i will assume three light neutrinos .
recall , the observed small neutrino masses can naturally be generated via the see - saw mechanism .
the @xmath5 neutrino mass matrix is given by @xmath6 , where @xmath7 ( @xmath8 ) is the dirac rl ( majorana rr ) mass matrix .
of course , the main problem with neutrino mixing angles is the fact that they are significantly larger than ckm mixing .
these large mixing angles may be obtained from one ( or a combination ) of the following sources
@xmath7 , @xmath8 ; or @xmath9 ( the charged lepton mass matrix ) . in the models presented below
, the origin of large neutrino mixing angles is in @xmath7 .
consider now the fgy @xcite ansatz given by @xmath10 where @xmath11 are the two right - handed ( sterile ) neutrinos and @xmath12 this neutrino mass matrix ansatz is expressed in the lepton flavor eigenbasis .
the dimensionful parameters @xmath13 are chosen to satisfy the relations @xmath14 for maximal atmospheric mixing angle and @xmath15 for a large , but not maximal , solar mixing angle .
in addition , due to the zeros in the mass matrix ansatz , there is only one non - vanishing cp violating angle , @xmath16 . upon integrating out the heavy sterile neutrinos ( @xmath17 )
, we obtain the @xmath5 fgy light neutrino mass matrix given by @xmath18 the neutrino mass eigenvalues and small mixing angle are given by @xmath19 a detailed @xmath20 analysis of the fgy ansatz including atmospheric and solar neutrino oscillation data was performed by raidal and strumia @xcite .
their best fit then makes predictions for @xmath21 , observable in planned long baseline experiments , and for the effective electron neutrino mass measured in neutrinoless double beta decay , @xmath22 mev , which is unobservable in any planned experiment .
in addition , rs perform a detailed analysis of leptogenesis with fgy .
they find two successful solutions providing an acceptable cosmological baryon asymmetry [ _ this is the rs ambiguity _ ] .
one with @xmath23 in this case , the prediction for cp violating neutrino oscillation is given by @xmath24 in a supersymmetric generalization of fgy , they predict the lepton number flavor violating branching ratios @xmath25 where @xmath26 . the other with @xmath27 they find @xmath28 and @xmath29
we now obtain the fgy texture in a supersymmtric theory with an [ su(2)@xmath30u(1)]@xmath31 family symmetry protecting the zeros @xcite .
this is a theory of leptons with the 3 families of electroweak doublets given by @xmath32 .
the first two families transform as a doublet under the su(2 ) family symmetry with @xmath33 , and then @xmath34 is an su(2 ) singlet .
the superpotential for the neutrino mass sector is given by ( we will discuss the charged lepton mass sector next ) @xmath35 where @xmath36 is a new scale satisfying @xmath37 .
the familon fields @xmath38 are assumed to get the following vacuum expectation values [ vevs ] breaking the family symmetry and generating neutrino masses .
we have , in complete generality , @xmath39 and @xmath40 .
in addition the two right - handed neutrinos @xmath41 obtain majorana masses when the su(2 ) singlet familon fields get vevs given by @xmath42 .
the family symmetry [ with specific u(1 ) charges for all the fields , as discussed in more detail in the paper @xcite ] is sufficient to make this the most general superpotential consistent with the symmetry .
hence the zeros of the fgy ansatz are obtained , without fine tuning .
when the higgs doublet @xmath43 gets a vev at the weak scale given by @xmath44 we finally obtain the fgy neutrino mass matrix with the parameters @xmath45 given below . @xmath46 note , however , unlike fgy we must now consider the charged lepton mass matrix which is also constrained by the family symmetry . until we do this , we can not be certain that we have obtained the fgy ansatz in the lepton flavor basis .
consider the superpotential for charged leptons given by @xmath47 where the left - handed anti - leptons @xmath48 are su(2 ) singlets .
when the higgs @xmath49 gets a vev given by @xmath50 we obtain the charged lepton @xmath5 mass matrix below @xmath51 with @xmath52 satisfying @xmath53 .
note the parameters @xmath54 are , up to order one coefficients , the same as @xmath45 appearing in the neutrino mass matrix .
the charged lepton mass eigenvalues are approximately given by @xmath55 and the charged lepton mass matrix is diagonalized by the bi - unitary transformation @xmath56 with the unitary matrix defining the left - handed mass eigenstates satisfying @xmath57 .
using this matrix we finally obtain the neutrino mass matrix in the flavor basis .
we find @xmath58 \ ; u_e \approx { \cal m}_{fgy } & \end{aligned}\ ] ] with @xmath59 note , the ratios @xmath60 and @xmath61 can be adjusted to accommodate bi - large neutrino mixing . as a bonus
we now see that the ratio of majorana neutrino masses @xmath62 is fixed .
we find @xmath63 .
thus @xmath64 and the rs ambiguity is resolved !
note , an su(2 ) family symmetry in susy theories is desirable for completely different reasons . it has been shown that it can ameliorate the susy flavor problem @xcite .
for example , prior to su(2 ) symmetry breaking the first and second generation sleptons are degenerate .
hence the off diagonal smuon - selectron mass term necessary for processes such as @xmath65 , seen in the figure below , are suppressed .
( 300,70)(0,0 ) ( 40,30)(130,30 ) ( 35,40)[rc]@xmath66 ( 130,30)(30,180,0 ) ( 155,45)(210,70)42 ( 220,60)[rc]@xmath67 ( 130,60)[lc]x ( 155,65)[rc]@xmath68 ( 113,64)[rc]@xmath69 ( 130,30)(220,30 ) ( 225,40)[lc]@xmath70 finally , at the moment we have an effective higher dimensional field theory with new scales @xmath71 . recall that successful leptogenesis requires @xmath72 gev .
hence it is reasonable to expect that @xmath73 gev . consider the following virtues of susy guts .
we have * @xmath74 naturally " * explains charge quantization * predicts gauge coupling unification@xmath75 * predicts yukawa coupling unification * + family symmetry @xmath76 hierarchy of fermion masses and protects against large flavor violation * neutrino masses via see - saw scale @xmath77 @xmath78 * lsp dark matter candidate , and * baryogenesis via leptogenesis with all of these virtues it is worth considering embedding our generalized mssm into an so(10 ) susy gut .
we consider the following so(10 ) susy gut with an @xmath79_{fs}$ ] proposed intially in @xcite and analyzed in great detail in @xcite .
the superpotential for the charged fermion sector , including the heavy froggatt - nielsen [ fg ] states \ { @xmath80 , @xmath81 } , familon fields \ { @xmath82 } and
so(10 ) adjoint \ { 45 } , is given by @xmath83 where we have the three families in @xmath84 ; and the higgs doublets in @xmath85 .
in addition , the fg mass @xmath86 necessarily includes so(10 ) breaking vevs with @xmath87 .
@xmath88 are the so(10 ) breaking vevs in the adjoint representation of so(10 ) with @xmath89 corresponding to the u(1 ) in so(10 ) which preserves su(5 ) , and @xmath90 the standard weak hypercharge .
@xmath91 are arbitrary parameters . upon integrating out the fg fields we obtain the effective fermion mass operators in the figure . finally upon giving the familon fields vevs we obtain the effective yukawa couplings below .
@xmath92 with @xmath93 the small parameters are given by @xmath94 ; all other parameters are numbers of order one .
all these parameters are fit to the low energy data via a detailed @xmath20 analysis .
the model has the following nice features .
the family hierarchy is obtained via the hierarchy of family symmetry breaking @xmath105 where the first ( second ) breaking is due to the familon vevs determining the small ratios @xmath106 ( @xmath107 ) . as a result
the 3rd family is much heavier than the 2nd family ; is then heavier than the 1st family .
we incorporate the following mass patterns 1 ) an approximate georgi - jarlskog mechanism naturally " with the familon vev @xmath108 giving @xmath109 and @xmath110 ; 2 ) third generation yukawa unification with @xmath111 @xmath112 @xmath113 and 3 ) @xmath114 leads to @xmath115 even though @xmath116 .
the model has 10 yukawa parameters ( 6 real coefficients and 4 phases ) to fit 13 fermion masses and mixing angles .
varying these 10 yukawa parameters [ three gauge ( @xmath118 ) and 7 soft susy breaking parameters ] at @xmath119 we use the two ( one ) loop rg equations for dimensionless ( dimensionful ) variables to obtain the @xmath20 function at @xmath120 .
all observables at @xmath120 are evaluated including one loop threshold corrections .
the result of the @xmath20 analysis , using the code of t. blazek ( see refs .
@xcite ) , is given in the table .
the fit is quite good .
in particular there are only 10 yukawa parameters with 16 independent fermion mass and mixing angle observables .
let us now consider neutrino masses in this model .
the model has three right - handed neutrinos contained in the fields @xmath123 .
moreover the dirac neutrino yukawa matrix is given by @xmath124 with @xmath125 this then gives the dirac lr neutrino mass matrix @xmath126 .
the field @xmath129 ( and another field @xmath130 ) are assumed to obtain vevs in the right - handed neutrino direction .
in fact , these vevs are necessary ( along with the @xmath131 vevs ) to break so(10 ) to the standard model .
given @xmath132 , @xmath133 and the familon vevs we obtain the effective neutrino mass terms given by @xmath134 where @xmath135 note , defining the @xmath139 dirac neutrino mass matrix @xmath140 and the @xmath141 right - handed neutrino mass matrix @xmath142 we finally obtain the light neutrino mass matrix @xmath143 \ ; u_e .\ ] ] hence we obtain the frampton , glashow & yanagida ansatz without any fine - tuning . in particular , the zeros are exact and fixed by the family symmetry
. this may be seen by the intermediate calculation below .
in addition , bi - large mixing requires significant constraints on the parameters @xmath145 .
maximal atmospheric neutrino oscillation requires @xmath146 , whereas large mixing for solar neutrino oscillations requires @xmath147 of order @xmath148 . on the other hand ,
the ratios @xmath61 and @xmath60 are , now , also constrained by charged fermion masses and mixing angles . without performing a joint @xmath20 analysis ( including both charged fermions and neutrinos ) we can still make the following observations .
the fit to charged fermion masses naturally gives @xmath149 .
hence given @xmath150 we find @xmath151 . with regards
@xmath152 we have @xmath153
. moreover @xmath154 gives @xmath155 which apparently requires fine - tuning of order 1 in 10 to get @xmath15 .
clearly in order to test this model , we must perform an extended @xmath20 analysis including both charged fermions and neutrinos .
note , there are only two new parameters in the neutrino sector which can be fit to the solar and atmospheric neutrino mass squared differences . finally we find @xmath156 and thus @xmath157 .
unfortunately , leptogenesis is more complicated in this model since we must first identify the heavy neutrino mass eigenstates and their lepton / higgs couplings .
moreover there are , in principle , more cp violating phases . the frampton - glashow - yanagida neutrino mass matrix ansatz has several nice features .
since it has only one cp violating angle it can , in principle , correlate the sign of the matter - anti - matter asymmetry with cp violating neutrino oscillations .
unfortunately , the raidal - strumia analysis uncovered an ambiguity which allows either sign of the cp violating angle for successful leptogenesis , depending on whether the right - handed neutrino masses satisfies @xmath158 or @xmath159 !
in addition the fgy ansatz provides a natural framework for bi - large neutrino mixing ! in this talk we obtained the fgy ansatz in a supersymmetric model with an [ su(2 ) @xmath30 u(1)]@xmath31 family symmetry .
this had two advantages , 1 ) the fgy ansatz is fixed by the family symmetry and 2 ) it resolves the rs ambiguity .
this is because the same family symmetry acting on neutrinos also constrains the charged fermion masses . as a result ,
the ratio of right - handed neutrino masses is fixed by the ratio of @xmath160 for atmospheric and solar neutrinos and the ratio of charged fermion masses .
it is also important to note that we are only able to obtain the normal hierarchy for neutrino masses .
the second model presented in this talk was a susy so(10 ) model with an [ su(2 ) @xmath30 u(1)@xmath161@xmath31 family symmetry .
this model is significantly more constrained since it relates all fermion masses and mixing angles .
there are only two new arbitrary parameters in the neutrino sector which are fixed by the neutrino mass differences .
nevertheless , we naturally " obtain the fgy ansatz for neutrinos ( in a theory which initially starts with 3 right - handed neutrinos and 3 so(10 ) singlet neutrinos ) .
we find a large neutrino mixing angle for atmospheric neutrino oscillations naturally . " on the other hand , an extended @xmath20 analysis must be performed to see if we can obtain a large solar mixing angle .
if a good fit is obtained for neutrino masses and mixing , it will then be very interesting to analyze cp violation and leptogenesis in this model . 9 p. h. frampton , s. l. glashow and t. yanagida , phys .
b * 548 * , 119 ( 2002 ) [ arxiv : hep - ph/0208157 ] .
m. raidal and a. strumia , phys .
b * 553 * , 72 ( 2003 ) [ arxiv : hep - ph/0210021 ] .
s. raby , phys .
b * 561 * , 119 ( 2003 ) [ arxiv : hep - ph/0302027 ] .
m. dine , r. g. leigh and a. kagan , phys .
d * 48 * , 4269 ( 1993 ) [ arxiv : hep - ph/9304299 ] ; a. pomarol and d. tommasini , nucl .
b * 466 * , 3 ( 1996 ) [ arxiv : hep - ph/9507462 ] ; r. barbieri , g. r. dvali and l. j. hall , phys .
b * 377 * , 76 ( 1996 ) [ arxiv : hep - ph/9512388 ] .
r. barbieri , l. j. hall , s. raby and a. romanino , nucl .
b * 493 * , 3 ( 1997 ) [ arxiv : hep - ph/9610449 ] .
t. blazek , s. raby and k. tobe , phys .
d * 60 * , 113001 ( 1999 ) [ arxiv : hep - ph/9903340 ] ; t. blazek , s. raby and k. tobe , phys .
d * 62 * , 055001 ( 2000 ) [ arxiv : hep - ph/9912482 ] .
t. blazek , m. carena , s. raby and c. e. m. wagner , phys .
d * 56 * , 6919 ( 1997 ) [ arxiv : hep - ph/9611217 ] .
|
in this talk i describe a natural framework for bi - large neutrino mixing within the context of two models 1 ) a simple generalization of the mssm and 2 ) an so(10 ) model .
our starting point is the frampton , glashow , yanagida [ fgy ] neutrino mass ansatz which can easily accomodate bi - large neutrino mixing .
the main point of fgy , however , is to obtain a theory of neutrino masses with only one possible cp violating angle .
they argue that the sign of the baryon asymmetry of the universe ( assuming leptogenesis ) is then correlated with cp asymmetries possibly observable in accelerator experiments .
unfortunately , there is a fly in the ointment .
it was later shown by raidal and strumia [ rs ] that there is a sign ambiguity which frustrates the above correlation .
we note that the raidal - strumia ambiguity is resolved in our models .
|
charmonium strong decays of the type @xmath30 , where @xmath1 is a generic charmonium state , @xmath2 is a nucleon and @xmath3 is a light meson , have recently attracted interest both as sources of information regarding the @xmath5 spectrum @xcite and in searches for a low energy @xmath31 enhancement @xmath32 " , which has been reported in @xmath33 @xcite and @xmath34 @xcite , but thus far not in @xmath35 .
these decays are also of interest because their partial widths can be used to estimate the @xmath36 associated charmonium production cross sections at panda @xcite .
as we shall show here , they may also provide information on @xmath37 meson - nucleon coupling constants , which could be used to identify unusual resonances such as molecule or glueball candidates .
specific @xmath30 reactions that have recently been studied experimentally include @xmath38 @xcite , @xmath39 and @xmath40 @xcite , and @xmath41 @xcite ; @xmath42 @xcite , @xmath43 @xcite , @xmath44 @xcite , @xmath45 @xcite and @xmath41 @xcite , and ( upper limit ) @xmath46 @xcite ; and @xmath47 and @xmath39 @xcite .
these decays may prove to be complicated processes in which several decay mechanisms contribute significantly .
for this reason it will be useful to have predictions for @xmath30 dalitz plot ( dp ) event densities assuming various decay mechanisms ; this paper provides these results for one such mechanism .
in particular we derive the dp event densities that follow from sequential meson emission , in which the charmonium state ( generically @xmath1 ) decays to an intermediate @xmath31 state , which radiates the light meson from the @xmath2 or @xmath48 line , @xmath49 .
the two feynman diagrams assumed in this model are shown in fig.[fig : blr_fig1 ] .
-4 mm -7 cm we emphasize that the actual relative importance of this and other @xmath30 decay mechanisms is unclear at present , and may depend strongly on the charmonium state @xmath1 and the light meson @xmath3 ; one purpose of this paper is to determine the rates predicted by this meson emission decay model in isolation for comparison with experiment , so that the importance of this decay mechanism can be estimated .
the predictions of this @xmath49 decay model can be given in some cases with no free parameters , since the strengths of the _ a priori _ unknown couplings @xmath50 and @xmath37 can be estimated from other processes .
here we will give absolute predictions for the set of partial widths @xmath51 ; we use the known partial widths @xmath52 to estimate the @xmath53 couplings , and the final @xmath54 coupling is of course well known .
provided that the contribution of the @xmath49 decay mechanism can be isolated and quantified experimentally , this information can be used to estimate meson - nucleon strong coupling constants ; these are generally poorly known , and play an important role in nuclear physics as input parameters in meson exchange models of the @xmath23 force @xcite .
we will also discuss the determination of the ( well known ) @xmath55 and the ( poorly known ) @xmath41 couplings from the decays @xmath38 , @xmath42 and @xmath56 as examples .
this provides a third motivation for the study of @xmath57 decays ; they may prove useful for estimating @xmath37 coupling constants , in addition to their relevance to @xmath5 spectroscopy @xcite and low - mass @xmath58 dynamics @xcite .
another motivation for studying @xmath30 is the possibility of observing light scalars , including the @xmath59 " , the 980 mev states and the scalar glueball , in the decays @xmath60 ( and @xmath61 ) .
here we will usually specialize to charmonium decays to a @xmath62 pair and a neutral meson , @xmath63 ; these decays are reasonably well studied , and enjoy the simplification of equal baryon and antibaryon masses .
our results employ conventions for kinematic variables , meson - baryon couplings and masses that were used in ref.@xcite .
in particular , @xmath64 is the mass of the initial charmonium state , @xmath65 is the proton mass , @xmath66 is the mass of meson ( subscript ) @xmath3 , and dimensionless mass ratios @xmath67 and @xmath68 are defined relative to the proton mass .
( hence the numerical values of @xmath69 and @xmath70 depend on the decay process . )
we use scaled dimensionless variables @xmath71 and @xmath72 and their inverses @xmath73 and @xmath74 to describe dalitz plots ; these greatly simplify our results .
the dp event densities we present here are formally partial width densities in @xmath75 and @xmath76 , which are related to the more familiar differential partial widths by a trivial overall constant , = m_p^4 .
[ eq : dp_densities_defn ] before we give our results for these event densities , it is useful to recall some general properties of a @xmath35 dalitz plot .
the boundary in the dimensionless variables @xmath77 is specified by the curves y _ = [ eq : dp_boundaries ] where @xmath78 and @xmath79 , with f(a , x ) ( a^2 ( a^2 -4 ) - 2a^2 x + x^2)^1/2 .
[ eq : dp_boundaries_suppl ] the range of values of @xmath75 ( and @xmath76 ) in the physical region is r ( r+2)x r ( r-2 ) . [
eq : dp_xrange ] the areas @xmath80 of these dalitz plots are useful for estimating @xmath81 partial widths @xcite .
although @xmath82 can be evaluated in closed form for @xmath35 with general mass ratios @xmath70 and @xmath69 , the resulting expression is quite lengthy , so when required we will simply evaluate each @xmath82 numerically .
in deriving the dp event densities we have usually assumed that the @xmath83 coupling is a constant @xmath84 times the simplest relevant dirac matrix for the given @xmath1 quantum numbers ; for example , for the @xmath7 we use a pure vector @xmath85 vertex , @xmath86 .
the order of the hadron labels in @xmath87 is meant to reflect the fact that the numerical value of this coupling constant is taken from an @xmath88 transition , here @xmath89 .
this could be a significant concern if form factor effects are large .
we proceed similarly for the light mesons @xmath90 and @xmath91 ; for the pion we use a pure pseudoscalar @xmath54 coupling , with vertex @xmath92 , and @xmath93 for the @xmath94 vertex .
since light vector mesons ( generically represented by the @xmath12 ) have two interesting strong couplings , dirac ( vector ) and pauli ( anomalous magnetic ) , for this special case we assume a vertex with two interactions , ^ ( ) _ = -ig_nn ( _ + i ( _ nn/2 ) _ q _ ) .
[ eq : omega_vertex ] ref.@xcite assumed a similar @xmath85 vertex ; see ref.@xcite for additional details regarding the couplings assumed here .
we generally abbreviate these coupling constants as @xmath95 and @xmath96 ; rationalized squared couplings @xmath97 and @xmath98 are also used .
for the special case of @xmath55 final states , these event densities can be obtained by applying crossing relations to our previously published results for the unpolarized differential cross sections for the @xmath99 processes @xmath100 @xcite ; the other ( @xmath91 and @xmath12 ) cases have not been considered previously .
the results for all cases considered here are given below .
& & = _ _ c _ \ { ( u - v)^2 + & & ( - r^2 r^2 ) } [ eq : etac_pppi0 ] + + & & = _ j/ _ \ { + & & - 2(u+v ) ( u+v+1 ) r^2 + 2uv r^4 - ( u^2+v^2 ) + & & r^2 r^2 } [ eq : jpsi_pppi0 ] + + & & = _ _ c0 _ \ { ( u+v)^2 + & & ( + 4 r^2 - r^2 r^2 ) } [ eq : chic0_pppi0 ] + + & & = _ _ c1 _ \ { - + & & ( u+v+1 ) + r^2 + r^2 + ( 2(u^2+v^2 ) + & & + u + v ) r^2 r^2 -uv r^4 r^2 - r^2 r^4 } [ eq : chic1_pppi0 ] + + & & = _ _ c _ f_0 \ { ( u+v)^2 + & & ( + 4 r^2 - r^2 r^2 ) } [ eq : etac_ppf0 ] + + & & = _ j/
{ + & & + 8 ( u+v)(u+v+1 ) -2 ( u(u+1 ) + v(v+1 ) + & & + 6uv ) r^2 + 4(u+v)^2 r^2 + 2uv r^4 - ( u^2+v^2 ) + & & r^2 r^2 } [ eq : jpsi_ppf0 ] + + & & = _ _ c0 _ f_0 \ { + & & - 16(u+v)(u+v+1 ) + 4 ( u+v)^2 ( r^2 + r^2 ) + & & - ( u - v)^2 r^2 r^2 } [ eq : chic0_ppf0 ] + + & & = _ _ c1 _ f_0 \ { - + & & ( u+v+1 ) + r^2 + ( - 8(u+v ) + & & ( u+v+1 ) ) r^2 + ( 2(u^2+v^2 ) + 8uv + u + v ) r^2 r^2 + & & + 2 ( u+v)^2 r^4 -uv r^4 r^2 - r^2 r^4 } [ eq : chic1_ppf0 ] + + & & = _ _ c _ \ { + & & + _ + & & + _ ^2 } [ eq : etac_ppw ] + + & & = _ j/ _ \ { + & & + _ + _ ^2 } [ eq : jpsi_ppw ] + + & & = _ _ c0 _ \ { + & & + _ + & & + _ ^2 } [ eq : chic0_ppw ] + + & & = _ _ c1 _ \ { + _ + _ ^2 } . [ eq : chic1_ppw ] the symmetry of these event densities under @xmath77 and hence @xmath101 interchange is a consequence of @xmath102-parity invariance .
there are singularities in these events densities along the lines @xmath103
( @xmath104 ) and @xmath105 ( @xmath106 ) , corresponding to @xmath107 and @xmath108 .
these are due to the poles of the @xmath109 and @xmath110 propagators in the feynman diagrams for the decay process @xmath111 ( fig.[fig : blr_fig1 ] ) and lie outside the physical regions of the dalitz plots .
as a first application we will evaluate the partial width for @xmath38 assuming this meson emission decay mechanism .
the pdg @xcite quotes a branching fraction of b(j/ ) = 1.09 0.09 10 ^ -3 [ eq : jpsi_pppi_bf_pdg ] which is the average of early measurements by mark - i @xcite , dasp @xcite and mark - ii @xcite .
there are also more recent experimental results on this decay from bes - ii @xcite . using the current pdg value for the @xmath7 total width of @xmath112 kev
, this branching fraction corresponds to a partial width of ( j/ ) = 102 9 .
[ eq : jpsi_pppi_width_pdg ] to evaluate this partial width in the meson emission model we simply integrate the theoretical event density ( [ eq : jpsi_pppi0 ] ) over the dalitz plot .
this event density is given by = _
j/ _ ( x , y ) [ eq : jpsi_pppirate ] where the dimensionless density function @xmath113 is ( x , y ) = \ { -2(u+v)(u+v+1 ) r^2 + 2 uv r^4 - ( u^2+v^2 ) r^2 r^2 } .
[ eq : rho_jpsi_pppirate ] this ( parameter - free ) relative event density , scaled to the maximum value in the physical region , is shown in fig.[fig : blr_fig2 ] .
-6 mm dp event density predicted by the meson emission decay mechanism @xmath114 , eq.([eq : jpsi_pppi0 ] ) .
contours of equal density are shown.,title="fig : " ] integrating ( [ eq : jpsi_pppirate ] ) over the dalitz plot gives ( j/ ) = _ j/ _ m_p a_d / m_p^4 .
[ eq:3body_rate ] where @xmath115 is the mean value of @xmath116 in the dalitz plot , which has ( physical ) area @xmath82 ; _ dp dx dy = a_d / m_p^4 .
[ eq : dp_int_defn ] we evaluate @xmath115 and @xmath82 numerically , assuming physical hadron masses ; we use pdg masses rounded to 0.1 mev ; @xmath117 gev , @xmath118 gev and @xmath119 gev , which leads to @xmath120 and @xmath121 gev@xmath122 , and a partial width of ( j/ ) = 34.44 _ j/ _ .
[ eq:3body_rate_num1 ] to complete this estimate we require numerical values for the @xmath54 and @xmath123 coupling constants . for @xmath54
there is general agreement from meson exchange models of @xmath23 scattering that @xmath124 ( see for example @xcite ) ; we accordingly set @xmath125 .
the value of the @xmath126 coupling constant ( here @xmath127 ) can be estimated from the measured partial width to @xmath58 , which is ( again using pdg numbers ) @xmath128 ev . equating this to the theoretical decay rate ( j / p|p ) = _ j/ _ p
( 1 + 2/r^2 ) m / 3 [ eq : jpsi_ppbarwidth_formula ] gives a value of @xmath129 , as was quoted previously in ref.@xcite .
using these couplings in eq.([eq:3body_rate_num1 ] ) gives our meson emission model prediction ( j/ ) = 97 .
[ eq:3body_rate_num2 ] this is consistent with the experimental value of @xmath130 ev quoted in eq.([eq : jpsi_pppi_width_pdg ] ) .
this excellent agreement is somewhat fortuitous , since this version of the model does not include the @xmath5 contributions evident in the @xmath38 dalitz plot @xcite ( see also fig.[fig : blr_fig3 ] ) .
we note in passing that the charged - pion cases @xmath131 and @xmath132 should have branching fractions close to twice @xmath133 , reflecting an isospin factor of two . experimentally this is indeed the case ; the ratio of the pdg @xmath7 branching fractions to @xmath134 and @xmath55 is @xmath135 .
projections of dp event densities onto the invariant mass of one pair of particles are useful in searches for intermediate resonances .
for @xmath38 , the bes collaboration has published acceptance - corrected event densities in @xmath19 and @xmath136 invariant mass ( fig.6 of ref.@xcite ) , which show clear evidence for @xmath5 resonances . here
we will generate the corresponding theoretical @xmath19-projected event distributions in the meson emission model for comparison with experiment .
although @xmath5 resonances are not incorporated in our calculation , this comparison will test the relative importance of the meson emission decay mechanism in this decay , and establish whether the model predicts a non-@xmath5 background " invariant mass distribution that is similar to the data in form and magnitude .
the full two - dimensional dp event density @xmath137 predicted by the meson emission model is given by eq.([eq : jpsi_pppi0 ] ) .
projecting this onto @xmath19 is straightforward ; first one integrates over @xmath138 between the dp boundaries @xmath139 of eq.([eq : dp_boundaries ] ) , which gives @xmath140 .
converting this into a distribution in @xmath19 introduces a jacobean , which is specified by the definition @xmath141 .
this gives = ^y_+(x)_y_-(x ) dy .
[ eq : projected_mass_distn ] we have evaluated this distribution numerically , given the @xmath142 of eq.([eq : jpsi_pppi0 ] ) and the masses and couplings @xmath143 and @xmath144 used in sec.iii.a .
the result is shown in fig.[fig : blr_fig3 ] , together with an experimental distribution provided by bes ( reported in ref.@xcite ) , using a common scale .
the data is the combined acceptance - corrected @xmath19 and @xmath136 distribution , scaled to give their reported @xmath145 rather than the pdg value of @xmath146 .
experimental ( bes ) and theoretical ( meson emission model , fig.[fig : blr_fig2 ] ) dalitz plot ( dp ) event densities , projected onto @xmath19 .
this is not a fit ; see text for discussion . ]
clearly there is a close resemblance between the meson emission model prediction for the @xmath38 event distribution in @xmath19 and the observed bes distribution , both in form and magnitude .
this suggests that a study of this reaction assuming this model for the experimental background " combined with @xmath5 resonance contributions would be an interesting exercise . although bes @xcite recently reported a similar study , they introduced an _ ad hoc _ @xmath147-dependent form factor that suppressed this background " meson emission amplitude relative to @xmath5 contributions .
the similarity to experiment evident in fig.[fig : blr_fig3 ] suggests that this mechanism merits additional consideration .
since the two meson emission model parameters @xmath84 and @xmath149 are both known for several @xmath148 decays , we are able to give absolute predictions for these partial widths .
( we previously used @xmath150 to estimate @xmath84 @xcite ; here we again use these values , and set @xmath151 . ) these @xmath148 partial widths are given in table [ table : pppi_rates ] , together with some intermediate theoretical quantities and the experimental widths .
( these experimental values are the pdg total widths times branching fractions , with errors added in quadrature . )
.comparison of theory ( meson emission model ) and experiment for @xmath152 ( see text ) . [ cols="^,^,^,^,>,>",options="header " , ] -0.5 cm we have also estimated the effect of an @xmath153 width on these results . imposing a breit - wigner @xmath91 mass profile with @xmath154 gev , truncated at @xmath155 , decreases all the absolute theoretical @xmath156 branching fractions in table [ table : ppf0_rates ] ( col.3 ) by @xmath157 .
the _ relative _ theoretical branching fractions ( col.2 ) are even less sensitive to the @xmath153 width , and become @xmath158 and @xmath159 .
a light scalar @xmath91 meson would presumably decay strongly and perhaps dominantly to @xmath160 , so decays of the type @xmath161 are of special interest , notably @xmath162 ( in view of our large theoretical @xmath163 ) .
the charged case @xmath164 has been studied by mark - i @xcite , desy @xcite and mark - ii @xcite . although no light scalar mesons have yet been identified in this decay , it is suggestive that @xmath164 is the largest known exclusive @xmath165 mode , with a branching fraction of @xmath166 .
in addition to the @xmath167 intermediate state of interest here , this decay may also receive important contributions from @xmath168 , @xmath169 and @xmath170 , as well as other two - baryon and @xmath37 states ; this may complicate the comparison with experiment considerably .
ref.@xcite , which has the largest event sample , finds a large but not dominant @xmath170 contribution , @xmath171 , and gives a rather tight upper limit of @xmath172 on the contributing subprocess @xmath173 ; @xmath174 ( 90% c.l . ) . as
ref.@xcite shows in their fig.31 that the @xmath175 invariant mass distribution from non-@xmath170 events is a broad sigma - like distribution , there may well be a large @xmath176 contribution to this decay , with a branching fraction comparable to the theoretical @xmath177 predicted for @xmath178 ( see table [ table : ppf0_rates ] ) .
it will be very interesting to investigate this possible light @xmath91 contribution in a future experimental study , as well as to search for the @xmath179 and @xmath180 scalar states and the scalar glueball candidate @xmath181 in ( higher - mass ) charmonium decays , notably of the @xmath10 .
charmonium decays to @xmath184 are especially interesting , since the @xmath12 plays a crucial role in meson - exchange models of the @xmath23 force , as the dominant origin of the short - ranged hard core repulsion " , through @xmath185-channel @xmath12 exchange .
conceptual problems with this @xmath12-exchange mechanism include 1 ) the very small @xmath23 separation implied by this mechanism ( @xmath186 fm ) , at which quark - gluon dynamics may be a more appropriate description of the interaction , and 2 ) the prediction of a corresponding short - ranged @xmath4 attraction and deeply bound @xmath4 states , which are not observed .
( see for example refs.@xcite , and references cited therein . )
there are two strong coupling constants in the @xmath187 vertex , as summarized by eq.([eq : omega_vertex ] ) , the overall strength @xmath188 of the dirac ( @xmath189 ) coupling , and the relative strong magnetic pauli coupling @xmath190 ( here abbreviated @xmath191 and @xmath192 , with @xmath193 ) .
unfortunately the @xmath183 couplings in the meson exchange models are not _ a priori _ well established experimentally , and are therefore usually fitted directly to @xmath23 scattering data .
these @xmath23 scattering studies are thus fits to the data rather than predictions that test the assumed vector - meson - exchange scattering mechanism .
these @xmath23 fits typically find @xmath194-@xmath195 for the dirac @xmath182 coupling , whereas the @xmath182 pauli coupling has remained poorly determined ; examples of @xmath182 parameter sets from the @xmath23 scattering literature include @xmath196 ( paris ) , @xmath197 ( nijmegen ) , and @xmath198 ( cd - bonn ) ( these are cited in ref.@xcite ) .
independent estimates of the @xmath182 coupling from experiment have been reported by sato and lee @xcite ( from pion photoproduction ) and by mergell , meissner and drechsel @xcite ( from nucleon em form factors ) .
et al . _ assumed @xmath199 , and quoted the range @xmath200-@xmath201 for experimentally favored values of the dirac coupling .
et al . _ found a small @xmath192 but a much larger dirac coupling , @xmath202 .
theoretical calculations include a qcd sum rule result of zhu @xcite , who finds @xmath203 , and a recent p0 quark model calculation of @xmath37 couplings @xcite which found the analytic result @xmath204 .
charmonium decays to @xmath41 final states ( and @xmath205 more generally ) may allow independent estimates of these @xmath182 and @xmath183 couplings , again provided that they are dominated by the meson emission decay mechanism , or at least that this contribution to the decay amplitude can be isolated and quantified . in the following discussion
we will consider the decay @xmath56 as an example .
results from experimental studies of the decay @xmath56 have been published by mark - i @xcite , mark - ii @xcite and most recently bes - ii @xcite .
the pdg value of the @xmath56 branching fraction , estimated from these results , is @xmath206 , which combined with the pdg @xmath7 total width gives an experimental partial width of ( j/ ) = 103 14 .
[ eq : jpsi_ppw_width_pdg ] the fact that this is approximately equal to the @xmath55 partial width ( [ eq : jpsi_pppi_width_pdg ] ) despite the much smaller phase space suggests a robust @xmath187 coupling .
on evaluating this theoretical decay rate by integrating eq.([eq : jpsi_ppw ] ) numerically with physical masses , both @xmath182 couplings free , and ( as used previously ) @xmath207 , we find ( j/ ) = ( 2.468 - 1.101 + 0.886 ^2 ) .
[ eq : jpsi_ppw_width_thy ] the single number @xmath208 alone does not suffice to determine the @xmath187 strong couplings since there are two free parameters , @xmath191 and @xmath192 .
if we set @xmath199 , following cd - bonn @xcite and the sato - lee photoproduction study @xcite , the measured partial width ( [ eq : jpsi_ppw_width_pdg ] ) and the theoretical decay rate ( [ eq : jpsi_ppw_width_thy ] ) imply @xmath209 ( provided that meson emission dominates this decay ) .
this @xmath191 is rather larger than these references prefer for @xmath191 ; it is consistent however with the em form factor fit of mergell _ et al . _
@xcite and the range @xmath210 reported by zhu @xcite from qcd sum rules .
if we instead assume the p0 quark model value @xmath204 for the pauli term @xcite , we find @xmath211 , which is consistent with the values quoted by @xmath23 scattering studies , and is somewhat larger than the photoproduction value . -6 mm dp event density for @xmath199 , from eq.([eq : jpsi_ppw ] ) .
note the suppression near @xmath62 threshold ( upper right).,title="fig : " ] it is possible to estimate the two parameters @xmath191 and @xmath192 independently through a more detailed comparison between @xmath56 data and the theoretical dp event density , eq.([eq : jpsi_ppw ] ) .
this theoretical event density is strongly dependent on the pauli coefficient @xmath192 ; with @xmath212 the event density at lower @xmath213 is strongly suppressed ( see fig.[fig : blr_fig4 ] ) .
in contrast , for moderately large @xmath214 , such as the quark model value @xmath204 , the theoretical dp event density is closer to uniform .
this is illustrated in fig.[fig : blr_fig5 ] , which shows this event density along the diagonal @xmath215 ( relative to the maximum value on diagonal ) for various @xmath216 .
if the meson emission model does give a good description of this decay , evidently it may be possible to determine @xmath192 by comparing the @xmath217 dp event density on diagonal to the prediction in fig.[fig : blr_fig5 ] .
dp event density along the diagonal @xmath215 , showing the strong @xmath192 dependence . ]
other decays to @xmath219 final states that are closely related to @xmath56 include @xmath220 and @xmath221 . in the meson emission model
these are both described by the decay formula ( [ eq : jpsi_ppw ] ) , albeit with different vector meson masses and @xmath183 couplings . given the rounded pdg masses @xmath222 gev and @xmath223 gev , we predict the numerical decay widths ( j/0 ) = ( 2.614 - 1.155 + 0.930 ^2 ) [ eq : jpsi_pprho_width_thy ] and ( j/ ) = ( 0.184 - 0.109 + 0.087 ^2 ) .
[ eq : jpsi_ppphi_width_thy ] the @xmath224 case is especially interesting due to the range of values reported for @xmath225 , as discussed by brown and machleidt @xcite .
although vector dominance predicts @xmath226 weak @xmath113 " , and some data has been interpreted as supporting this , fits to @xmath227 and s - d mixing in @xmath23 scattering prefer a larger value
strong @xmath113 " ; the bonn potential model for example uses @xmath228 @xcite .
a qcd sum rule calculation by zhu @xcite finds @xmath229 , comparable to the fitted bonn values .
in contrast , the valence quark model with a p0 @xmath230 coupling predicts a much smaller @xmath231 @xcite .
using eq.[eq : jpsi_pprho_width_thy ] we can give meson emission model predictions for @xmath232 that follow from these various @xmath233 parameters .
the bonn parameters cited above give @xmath234 and @xmath235 ; this is essentially equal to the current experimental upper limit @xcite of @xmath236 , which is a mark - ii result dating from 1984 .
the zhu qcd sum rule central values for @xmath233 give essentially identical results .
in contrast the valence quark model with p0 couplings @xmath237 and @xmath238 ( and using @xmath239 from the @xmath56 discussion above ) gives a much lower @xmath240 and hence @xmath241 , which is about a factor of 5 below the current experimental limit .
the proximity of the bonn and qcd sum rule parameter predictions for @xmath242 to the current limit suggests that an experimental study with significantly improved sensitivity could make a useful contribution to establishing @xmath230 couplings .
the decay @xmath221 in contrast _ has _ been observed , and has an experimental ( pdg ) branching fraction of @xmath243 , corresponding to @xmath244 .
unfortunately in this case we do not have a theoretical estimate for either @xmath245 or @xmath246 , since the ( valence level , leading - order ) p0 model predicts no @xmath247 coupling .
clearly it would be very interesting to obtain experimental values for these couplings , since little is known about the properties of zweig - suppressed vertices .
again , if the meson emission model gives a good description of this decay , a comparison of eq.([eq : jpsi_ppw ] ) to the experimental @xmath248 dp event distribution should allow an experimental determination of the @xmath247 couplings .
finally , we note in passing that @xmath10 decays to @xmath183 are apparently not in agreement with the meson emission model ; proceeding as above , we would predict a branching fraction of @xmath249 , whereas the pdg experimental value is an order of magnitude smaller , @xmath250 .
possible explanations for this discrepancy , including form factors and ( destructive interference with ) additional decay mechanisms , are discussed in the next section . since
the total @xmath251 data sample reported by cleo @xcite and bes @xcite comprises only about 35 events , it is not yet possible to establish the reason for this large discrepancy between experiment and the meson emission model .
this would ideally involve a comparison between the predicted and observed dalitz plot event densities .
hopefully this comparison will be possible using the large @xmath10 data set being accumulated at bes - iii .
in this paper we have presented and developed a hadron - level meson emission model " of charmonium decays of the type @xmath30 , where @xmath1 is a generic charmonium resonance , @xmath2 is a nucleon and @xmath3 is a light meson .
the model assumes that the decays take place through meson emission from the nucleon or antinucleon line , as a hadronic
final state radiation " correction to a @xmath252 transition .
as the model is relatively simple , we are able to evaluate the predicted dp event densities for many experimentally accessible and measured processes ; in particular we give event densities for @xmath6 , @xmath7 ( and @xmath10 ) , @xmath8 , @xmath9 and @xmath10 and @xmath11 and @xmath12 , and implicitly all cases with the same @xmath253 quantum numbers .
we used the reaction @xmath38 as a test case with no free parameters ( the @xmath254 and @xmath54 couplings are known ) , and compared the meson emission model predictions for the projected event density in @xmath19 and the partial width @xmath255 to experiment ; the partial width is in good agreement , and the @xmath19 event density appears to describe the non-@xmath5 background " contribution to this reaction observed experimentally .
we also give predictions for @xmath152 for all @xmath1 cases considered here ; the overall trend of large and small widths and their approximate scale is reproduced by the model
. we also considered scalar and vector meson production .
we estimated @xmath256 branching fractions for a light @xmath153 , and noted that the @xmath7 and @xmath10 are favored for these studies , and the @xmath10 is favored for glueball searches . in vector production we considered @xmath56 in particular , and noted that a high statistics study of this reaction could be used to estimate the @xmath182 couplings ( @xmath191 and @xmath192 ) , which play a crucial role in meson exchange models of the @xmath23 force .
we showed that the @xmath56 dalitz plot event density is rather sensitive to the poorly known @xmath182 pauli coupling @xmath192 .
determination of meson - nucleon strong couplings is a potentially very interesting application of @xmath35 decays .
there are several theoretical developments that will be very important for future applications of this model .
one should incorporate @xmath5 resonances ; this is conceptually straightforward but may be technically complicated , as it will introduce many new and poorly known resonance coupling parameters and phases .
this development is of course crucial to describe the data in reactions such as @xmath38 , which clearly shows @xmath5 resonance peaks ( fig.[fig : blr_fig3 ] ) .
another important development is the substitution of plausible @xmath257 and @xmath258 hadron vertex form factors for the assumed coupling constants ; the difficulty here is that hadronic form factors are poorly known , and models such as p0 that predict form factors have not been adequately developed and tested .
another interesting generalization of the strong vertices assumed here would be to include a @xmath85 pauli coupling ; as we noted previously @xcite , this can explain the observed @xmath259 angular distribution .
finally , one should include other significant decay mechanisms , as they become apparent through high - statistics studies of experimental dalitz plots .
these additional mechanisms might include intermediate meson resonances @xmath260 that couple strongly to @xmath31 , as in @xmath261 ; if the @xmath260 resonances lie in the physical region , they will give rise to characteristic @xmath260 resonance bands in @xmath213 that could be identified and incorporated in a more complete decay model .
we are happy to acknowledge useful communications with r.mitchell , k.seth , m.shepherd , e.s.swanson and b.s.zou regarding this research , and shu - min li and xiao - yan shen in particular for contributing the bes data used in preparing fig.[fig : blr_fig3 ] .
we also gratefully acknowledge the support of the department of physics and astronomy of the university of tennessee , the physics division of oak ridge national laboratory , and the department of physics , the college of arts and sciences , and the office of research at florida state university .
this research was sponsored in part by the office of nuclear physics , u.s .
department of energy .
m. ablikim _ et al . _
[ bes collaboration ] , phys .
d * 80 * , 052004 ( 2009 ) [ arxiv:0905.1562 [ hep - ex ] ]
. m. ablikim _ et al . _
[ bes collaboration ] , phys .
d * 71 * , 072006 ( 2005 ) [ arxiv : hep - ex/0503030 ] .
m. ablikim _ et al . _
[ bes collaboration ] , phys .
d * 74 * , 012004 ( 2006 ) [ arxiv : hep - ex/0605031 ] .
m. ablikim _ et al . _
[ bes collaboration ] , arxiv:0902.3501 [ hep - ex ] .
m. ablikim _ et al . _
[ bes collaboration ] , eur .
j. c * 53 * , 15 ( 2008 ) [ arxiv:0710.5369 [ hep - ex ] ] .
j. z. bai _ et al . _
[ bes collaboration ] , phys .
lett . * 91 * , 022001 ( 2003 ) [ arxiv : hep - ex/0303006 ] .
m. ablikim _ et al .
_ [ bes collaboration ] , phys . rev .
lett . * 95 * , 262001 ( 2005 ) [ arxiv : hep - ex/0508025 ] .
y.huang , confirmation of proton - antiproton mass threshold enhancement structure and x(1835 ) at bes - iii " ( presentation at hadron 2009 ) .
r. a. briere _ et al .
_ [ cleo collaboration ] , phys . rev .
* 95 * , 062001 ( 2005 ) [ arxiv : hep - ex/0505101 ] .
j. z. bai _ et al . _
[ bes collaboration ] , phys .
d * 67 * , 052002 ( 2003 ) [ arxiv : hep - ex/0209080 ] .
s. b. athar _ et al . _
[ cleo collaboration ] , arxiv : hep - ex/0611032 .
r. machleidt , k. holinde and c. elster , phys .
* 149 * , 1 ( 1987 ) .
r. machleidt , phys .
c * 63 * , 024001 ( 2001 ) [ arxiv : nucl - th/0006014 ] .
m. m. nagels , t. a. rijken and j. j. de swart , phys .
d * 15 * , 2547 ( 1977 ) . m. m. nagels , t. a. rijken and j. j. de swart , phys .
d * 20 * , 1633 ( 1979 ) .
v. g. j. stoks , r. a. m. klomp , c. p. f. terheggen and j. j. de swart , phys .
c * 49 * , 2950 ( 1994 ) [ arxiv : nucl - th/9406039 ] .
w. n. cottingham , m. lacombe , b. loiseau , j. m. richard and r. vinh mau , phys .
d * 8 * , 800 ( 1973 ) .
m. lacombe , b. loiseau , j. m. richard , r. vinh mau , j. conte , p. pires and r. de tourreil , phys .
c * 21 * , 861 ( 1980 ) . c. downum , d. phil .
, `` _ low energy nucleon - nucleon interactions from the quark model with applications _ '' , oxford university , 2009 ( unpublished ) . s. capstick and n. isgur ,
d * 34 * , 2809 ( 1986 ) .
s. capstick and w. roberts , phys .
d * 58 * , 074011 ( 1998 ) [ arxiv : nucl - th/9804070 ] .
t. barnes and x. li , phys .
d * 75 * , 054018 ( 2007 ) [ arxiv : hep - ph/0611340 ] .
t. barnes , x. li and w. roberts , phys .
d * 77 * , 056001 ( 2008 ) [ arxiv:0709.4491 [ hep - ph ] ] . c. amsler et al .
( particle data group ) , phys .
b*667 * , 1 ( 2008 ) , and 2009 partial update for the 2010 edition .
i. peruzzi _ et al .
_ , phys . rev .
d * 17 * , 2901 ( 1978 ) .
r. brandelik _ et al . _
[ dasp collaboration ] , z. phys .
c * 1 * , 233 ( 1979 ) .
m. w. eaton _ et al .
_ , phys .
d * 29 * , 804 ( 1984 ) .
h. j. besch _ et al .
_ , z. phys .
c * 8 * , 1 ( 1981 ) .
k. maltman and n. isgur , phys .
lett . * 50 * , 1827 ( 1983 ) .
k. maltman and n. isgur , phys .
d * 29 * , 952 ( 1984 ) .
t. barnes , s. capstick , m. d. kovarik and e. s. swanson , phys .
c * 48 * , 539 ( 1993 ) [ arxiv : nucl - th/9302007 ] . c. downum , t. barnes , j. r. stone and e. s. swanson , phys .
b * 638 * , 455 ( 2006 ) [ arxiv : nucl - th/0603020 ] .
t. sato and t. s. h. lee , phys .
c * 54 * , 2660 ( 1996 ) [ arxiv : nucl - th/9606009 ] .
p. mergell , u. g. meissner and d. drechsel , nucl .
a * 596 * , 367 ( 1996 ) [ arxiv : hep - ph/9506375 ] .
shi - lin zhu , phys .
c * 59 * , 3455 ( 1999 ) [ arxiv : nucl - th/9810031 ] .
g. e. brown and r. machleidt , phys .
c * 50 * , 1731 ( 1994 ) .
shi - lin zhu , phys .
c * 59 * , 435 ( 1999 ) [ arxiv : nucl - th/9809032 ] .
|
in this paper we consider a sequential `` meson emission '' mechanism for charmonium decays of the type @xmath0 , where @xmath1 is a generic charmonium state , @xmath2 is a nucleon and @xmath3 is a light meson .
this decay mechanism , which may not be dominant in general , assumes that an @xmath4 pair is created during charmonium annihilation , and the light meson @xmath3 is emitted from the outgoing nucleon or antinucleon line .
a straightforward generalization of this model can incorporate intermediate @xmath5 resonances .
we derive dalitz plot event densities for the cases @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 and @xmath11 and @xmath12 ( and implicitly , any @xmath13 , @xmath14 or @xmath15 final light meson )
. it may be possible to separate the contribution of this decay mechanism to the full decay amplitude through characteristic event densities .
for the decay subset @xmath16 the two model parameters are known , so we are able to predict absolute numerical partial widths for @xmath17 . in the specific case @xmath18 the predicted partial width and @xmath19 event distribution
are intriguingly close to experiment .
we also consider the possibility of scalar meson and glueball searches in @xmath20 .
if the meson emission contributions to @xmath21 decays can be isolated and quantified , they can be used to estimate meson - nucleon strong couplings @xmath22 , which are typically poorly known , and are a crucial input in meson exchange models of the @xmath23 interaction .
the determination of @xmath24 from @xmath18 and the ( poorly known ) @xmath25 and the anomalous `` strong magnetic '' coupling @xmath26 from @xmath27 are considered as examples .
p 0p|p f_0 0p|p ^0 p0@xmath28p@xmath29
|
we are engaged in a large scale project to find additional emp stars in the halo of our galaxy by mining the database of the he s .
the normal procedures outlined by @xcite to isolate emp stars from the candidate lists produced by the he s were followed .
the candidates were vetted via moderate resolution spectroscopy at large telescopes to eliminate the numerous higher abundance interlopers .
most of the follow up spectra for the stars discussed here were obtained with the double spectrograph on the hale telescope at palomar mountain .
we intend to observe all candidates to the magnitude limit of the he s ( b @xmath6 17.5 ) in our fields ; with @xmath61600 moderate resolution spectra in hand from campaigns at palomar and at the las campanas observatory , observations are now complete in @xmath6990 deg@xmath7 , complete to b=16.5 in an additional @xmath6700 deg@xmath7 , and approaching completion in the remaining fields . these follow up spectra are used to determine an accurate measure of the metallicity of the star via a combination of strength of absorption in h@xmath8 ( determining @xmath9 ) and in the ca ii line at 3933 ( the kp index ) , which then determines [ fe / h ] .
the calibration between the strength of the indices and resulting metallicity [ fe / h](hes ) adopted by the he s is described in @xcite and is essentially identical to that used by the hk survey until recently ; the latest updates to the algorithm as used by the hk survey are described in @xcite .
we have found that there are systematic scale errors in the [ fe / h ] values determined by this algorithm for certain emp highly c - enhanced giants . as is shown in @xcite ,
these scale errors act to make certain c - stars appear more metal poor by a factor of @xmath610 than would be inferred from a detailed abundance analysis of their high dispersion spectra .
these problems appear to arise due to molecular absorption of ch and cn in the specific continuum bandpasses used to measure the kp and the h@xmath8 line indices from which [ fe / h ] is inferred .
this makes both of these lines appear weaker , and hence the inferred fe - abundance of such a c - star is underestimated .
this effect is largest for the cooler c - stars ( @xmath9@xmath65100 k ) with high c - enhancement and high n / c ratios .
normal - c giants and the warmer c - stars are not affected ; [ fe / h](hes ) and [ fe / h](hires ] are in good agreement for them .
we rely on the results of an extensive set of detailed abundance analyses of stars expected to have [ fe / h ] @xmath10 dex selected from the hamburg / eso survey to establish these claims .
we have obtained and analyzed spectra with hires @xcite at the keck i telescope of @xmath660 he s candidate emp stars .
@xcite will present abundance analyses for 14 of the 16 known carbon stars from this he s sample .
fifteen c - normal giants have been analyzed to date as well , while @xcite studied a large sample of candidate emp dwarfs from the he s .
the first consequence of these scale errors is that the fraction of carbon stars at extremely low metallicities has been substantially overestimated in several papers in the recent literature .
our operational definition of a c - star is one whose spectrum shows bands of c@xmath11 . if no c@xmath11 bands are detected , but [ c / fe ] @xmath12 dex , we denote a star to be c - enhanced .
the strength of the c@xmath11 bands will be a function of @xmath9 , @xmath13(c ) , and to a lesser extent , log(@xmath14 ) and [ fe / h ] .
also we denote stars with @xmath9 @xmath15 k as `` dwarfs '' , while all cooler stars are called `` giants '' .
even though the cooler c - stars comprise only a small fraction of the most metal poor stars , an underestimate of a factor of @xmath61 dex in their [ fe / h](hes ) will have significant effects .
[ fig_feh_vmk ] shows [ fe / h](hes ) versus v - k for a sample of 489 emp candidates from the he s with moderate resolution spectra from the double spectrograph at the hale telescope .
the known c - stars and the c - enhanced star with hires analyses from this sample are indicated . in the upper panel , the c -
rich stars are plotted at their [ fe / h](hes ) values , while in the lower panel they are plotted at their [ fe / h](hires ) as determined from detailed abundance analyses .
although at their nominal fe - metallicities the c - stars dominate the population of the giants below [ fe / h](hes ) @xmath16 dex , using the results from analysis of high resolution spectra in the lower panel the frequency of c - stars among the most metal poor emp stars is reduced by a factor of 2.5 . the c - star frequency for the palomar sample of he s giants with [ fe / h](hes ) @xmath17 dex is 7.4@xmath2% .
adding in the fraction of c - enhanced stars among giants with [ fe / h](hes ) @xmath17 dex we establish below of 6.5@xmath18% , one obtains a total fraction of c - rich stars with [ c / fe ] @xmath19 dex of 14@xmath4% among our he s emp sample with [ fe / h](hes ) @xmath17 dex .
the metallicity distribution function is very sharply declining among halo stars at the lowest fe - metallicities .
thus the systematic errors we have found in the calibration of the he s and by inference the hk metallicity scale of @xcite , at least until quite recently ( see rossi _ et al . _ 2005 ) will also lead directly to systematic overestimates of the number of emp stars and of the yield for emp stars by these two major surveys .
more details of this work can be found in @xcite .
( he s ) versus v - k for a sample of 489 emp candidates from the he s with moderate resolution spectra from the double spectrograph at the hale telescope ( filled circles , limited to stars with [ fe / h](hes ) @xmath20 dex ) .
the c - stars are from this sample are indicated by filled stars ; the c - enhanced star is shown as an open star . in the upper panel the c - stars are plotted at their [ fe / h](hes ) values , while in the lower panel , at their [ fe / h](hires ) values .
the c - normal dwarfs have also been shifted towards higher [ fe / h ] by 0.38 dex in the lower panel . a typical error for a emp giant with normal c
is shown for a single star in the upper panel .
the vertical dashed line separates the giants from the dwarfs.,width=326 ]
we next establish the distribution of c abundances and c / fe ratios for those stars that are not classified as c - stars .
if [ c / fe ] @xmath21 dex , we call such stars c - enhanced . fig . [ fig_ch_vmk ]
displays the gp ( g band of ch ) index as a function of v@xmath22k color for the palomar sample .
the entire set of known c - stars from our database is shown ; for the giants they are concentrated at the top of the distribution , having the strongest gp indices at each color . to determine the c abundances we use the predicted ch band strengths computed by m. briley described in @xcite .
these were calculated using a full spectral synthesis for the gp index defined in @xcite .
@xmath23 is used to define @xmath9 , adopting a mean value of e(b@xmath22v ) of 0.05 mag for the he s stars .
[ fig_cfe ] shows the resulting histogram of [ c / fe ] for the giants from this sample with [ fe / h ] @xmath24 dex .
the histogram suggests a peak near [ c / fe ] @xmath6 0.3 dex , somewhat higher than the normal value for unmixed very metal poor giants of [ c / fe ] = 0.0 dex given by @xcite .
there is a sharp drop towards higher c / fe ratios , with the c - stars occupying the upper limit of the range near [ c / fe ] @xmath6 + 2.0 dex . the distribution slowly decreases for c / fe ratios lower than the peak .
the fraction of giants that are c - enhanced but are not known c - stars is 6.5@xmath18% .
is shown for the giants from the p200 sample of candidate emp stars with @xmath25 [ fe / h](hes ) @xmath24 dex .
the known c - stars are denoted by solid fill .
the predictions of m. briley described in @xcite for the strength of the g band of ch based on synthetic spectra were used.,width=345 ]
to establish the results presented above , we have carried out a detailed abundance analysis using high dispersion spectra from hires at keck for a sample of 14 of the 16 carbon stars found in our database among candidate empstars selected from the he s . using a @xmath9 scale based on v - i , v - j and v - k colors , we find that the fe - metallicities for the cooler c - stars ( @xmath9 @xmath65100 k ) have been underestimated by a factor of @xmath610 by the standard he s survey tools . since we have not used the high resolution spectra themselves to determine @xmath9 or log(@xmath14 ) , the ionization equilibrium is a stringent test of our analysis and procedures , including the assumption of lte .
we obtain a mean for log@xmath13(fe : fe ii ) @xmath22 log@xmath13(fe : fe i ) of @xmath26 dex , with a 1@xmath27 rms scatter about the mean of 0.18 dex .
c - enhancement in these very metal poor c - stars appears to reach a maximum just below the solar @xmath13(c ) , and shows evidence of decreasing with decreasing @xmath9 ( increasing luminosity ) , presumably due to mixing and dredge - up of c - depleted material .
[ fig_ba_feh ] shows [ ba / c ] as a function of fe - metallicity for this sample of c and c - enhanced stars .
ten of the c - stars from the he s that we have analyzed show an enhancement of ba ( and of the other @xmath28-process neutron capture heavy elements ) approximately equal to that of c. the other four show [ ba / c ] @xmath29 dex , i.e. a strong c enhancement , with more normal heavy elements , as contrasted to enhancement of both c and the @xmath28-process elements in the majority of the c - stars . including 10 additional c - stars compiled from the literature , @xmath680% of the full sample of emp / vmp c - stars show highly enhanced ba , while @xmath620% has [ ba / c ] @xmath30 dex .
it is clear from both the very high enhancements of lead and the ba / eu ratios seen among these c - stars that the @xmath28-process is responsible for the enhancement of the heavy neutron - capture elements , when present . at first sight
, this behavior suggests that two distinct processes are required to produce the ba - enhanced c - stars and those that are not .
nucleosynthesis within an intermediate mass agb star can reproduce the first set of characteristics , but these c - stars are in general of too low a mass and are not sufficiently luminous or evolved to be agb stars .
we suggest instead that the observed star is the former secondary in a binary system across which mass transfer has occured . what about the @xmath620% of the c - stars without heavy element enhancements ?
we suggest that there is no need to resort to intrinsic production or any other additional mechanism , and that essentially _ all _ of these stars are the original secondary stars in mass transfer binary systems .
we ascribe the differing enhancement of the @xmath28-process elements from c - star to c - star within our sample to some dependence in the nucleosynthetic yields involving , for example , the initial [ fe / h ] and mass of the original primary star . at the lowest metallicities ,
@xcite predict that little or no @xmath28-process enhancement will occur at the ba peak as there will be so few fe - seed nuclei that they will essentially all end up as lead .
[ fig_ba_feh ] provides some support for this prediction . in our hypothesis ,
essentially all of these c - stars were once binaries .
assuming most have survived to date without being disrupted , they should still be binaries with ( invisible ) white dwarf companions .
the statistics of binary detection among very metal poor c - stars from the he s can not be evaluated from our sample as most of the stars have only been observed at a single epoch .
so we look instead at the sample of c - stars added from the literature , which are mostly from the hk survey and are in general brighter than the he s c - stars in our sample .
they have been known as interesting objects for timescales of several years to a decade , giving more opportunity for radial velocity monitoring .
four of these 11 c - stars are confirmed binaries with measured periods , consistent with the preliminary results of the @xmath31 monitoring program of @xcite for @xmath28-process enhanced c - stars .
although the sample is small , considering the lack of suitable long - term radial velocity monitoring programs , the length of the typical period , the small velocity amplitudes , the faintness of the stars , and the relatively short time they have been known to be interesting , we find our detection rate for binaries among very metal poor and emp c - strs to be consistent with all such stars being binaries ; simulations by @xcite support this .
there is as yet insufficient @xmath31 monitoring data for the much smaller sample of c - enhanced but not @xmath28-process enhanced stars .
@xcite will present full details .
is shown as a function of [ fe / h ] for a sample of 24 c - stars ( 14 from the present sample , denoted by filled stars , 10 from the literature , indicated by open circles ) and 2 c - enhanced dwarfs ( filled circles).,width=326 ] we are very greatful to m. briley for slightly extending his published grid of predicted ch indices for emp / vmp stars .
jgc and jm are grateful for partial support from nsf grant ast-0205951 .
jgc is grateful for support from the ernest fullam award of the dudley observatory which helped initiate this work .
the work of nc and fjz is supported by deutsche forschungsgemeinschaft ( grants ch 214/3 and re 353/44 ) .
nc acknowledges support through a henri chretien international research grant administered by the american astronomical society .
cohen , j. g. , christlieb , n. , mcwilliam , a. , shectman , s. , thompson , i. , wasserburg , g. j. , ivans , i. , dehn , karlsson & melendez , j. , 2004 , apj , 612 , 1107 cohen , j. g. , briley , m. & stetson , p. b. , 2005 , aj ( in press ) cohen , j. g. , _ et al .
_ , 2005b , in preparation cohen , j. g. , _ et al .
_ , 2005c , in preparation cohen , j. g. , _ et al .
_ , 2005d , in preparation lucatello , s. , tsagarides , s. , beers , t. c. , carretta , e. , gratton , r. g. & ryan , s. g. , 2005 , apj , 625 , 825
|
we demonstrate that there are systematic scale errors in the [ fe / h ] values determined by the hamburg / eso survey ( and by the hk survey by inference ) for certain extremely metal poor ( emp ) highly c - enhanced giants .
the consequences of these scale errors are that a ) the fraction of carbon stars at extremely low metallicities has been substantially overestimated in several papers in the recent literature b ) the number of emp stars known is somewhat lower than has been quoted in the recent literature c ) the yield for emp stars by the hk and the he s survey is somewhat lower than is stated in the recent literature .
a preliminary estimate for the frequency of carbon stars among the giants in the he s sample with @xmath0 [ fe / h ] @xmath1 dex is 7.4@xmath2% , and for c - rich giants with [ c / fe ] @xmath3 + 1.0 dex is 14.4@xmath4% .
we rely on the results of an extensive set of detailed abundance analyses of stars expected to have [ fe / h ] @xmath5 dex selected from the hamburg / eso survey to establish these claims .
these analyses of @xmath650 he s candidate extremely metal poor stars have been carried out in as homogeneous a manner as possible . here
we present the key results of detailed abundance analyses of 14 c - stars selected in this way about 80% of such c - stars show highly enhanced ba as well , with c enhanced by a factor of about 100 , and [ ba / c ] roughly solar .
these stars often show prominent lead lines , and presumably are the remnants of the secondary in a mass transfer binary system where the former primary was an agb star , which transferred substantial mass at that evolutionary stage .
the remaining 20% of the c - stars do not show an enhancemement of the s - process neutron capture elements around the ba peak .
they tend to be the most metal - poor stars studied .
we suggest that they too result from mass transfer across a binary system .
( published abstract will be shorter due to space limitations )
|
the study of @xmath0 decays plays an important role in understanding the origin of @xmath3 violation and physics of heavy flavor .
we expect that the parameters of the cabibbo - kobayashi - maskawa ( ckm ) matrix in the standard model , for instance , the three angles @xmath4 , @xmath5 and @xmath6 in the unitary triangle , can be well - determined from @xmath0 decays , especially from the charmless non - leptonic two - body b decays . experimentally , many @xmath0 experiment projects have been running ( cleo , babar , belle etc . ) , or will run in forthcoming years ( btev , cern lhcb , desy herab etc . ) . with the accumulation of the data ,
the theorists will be urged to gain a deeper sight into @xmath0 decays , and to reduce the theoretical errors in determining the ckm parameters from the experimental data . in the theoretical frame ,
the standard approach to deal with such decays is based on the low - energy effective hamiltonian which is obtained by the wilson operator product expansion method ( ope ) . in this effective hamiltonian
, the short - distance contributions from the scale above @xmath7 have been absorbed into the wilson coefficients with the perturbative theory and renormalization group method .
the wilson coefficients have been evaluated to next - to - leading order .
then the main task in studying non - leptonic two - body @xmath0 decays is to calculate the hadronic matrix elements of the effective operators .
however , we do not have a reliable approach to evaluate them from the first principles of qcd dynamics up to now .
generally , we must resort to the factorization assumption to calculate the hadronic matrix elements for non - leptonic @xmath0 decays , in which the hadronic matrix element of the effective operator ( in general , which is in the form of current - current four - quark operator ) can be approximated as a product of two single current hadronic matrix elements ; then it is parameterized into meson decay constant and meson - meson transition form factor .
the most popular factorization model is the bauer - stech - wirbel ( bsw ) model@xcite . in many cases ,
bsw model achieves great success , which can predict the branching ratios of many modes of non - leptonic @xmath0 decays in correct order of magnitude .
this factorization assumption does hold in the limit that the soft interactions in the initial and final states can be ignored .
it seems that the argument of color - transparency can give reasonable support to the above limit . because @xmath8 quark is heavy , the quarks from @xmath8 quark decay move so fast that a pair of quarks in a small color - singlet object decouple from the soft interactions .
but the shortcomings of this simple model are obvious .
first , the renormalization scheme and scale dependence in the hadronic matrix elements of the effective operators are apparently missed .
then the full decay amplitude predicted by bsw model remains dependent on the renormalization scheme and scale , which are mainly from wilson coefficients . in past years
, many researchers improved the simple factorization scheme and made many remarkable progresses , such as scheme and scale independent effective wilson coefficients@xcite , effective color number which is introduced to compensate the ` non - factorizable ' contributions , etc .
furthermore , some progresses in nonperturbative methods , such as lattice qcd , qcd sum rule etc.@xcite , allow us to compute many non - perturbative parameters in @xmath0 decays , such as the meson decay constants and meson - meson transition form factors .
every improvement allows us to have a closer look at the @xmath0 nonleptonic decays . except for the factorization approximation ,
another important approach has been applied to study many @xmath0 exclusive hadronic decay channels , such as @xmath9 , @xmath10 , @xmath11 etc .
this is pqcd method@xcite . in this method
, people assumes that @xmath0 exclusive hadronic decay is dominated by hard gluon exchange .
it is analogous to the framework of perturbative factorization for exclusive processes in qcd at large momentum transfer , such as the calculation of the electromagnetic form factor of the pion @xcite .
the decay amplitude for @xmath0 decay can be written as a convolution of a hard - scattering kernel with light - cone wave functions of the participating mesons .
furthermore , in ref.@xcite the sudakov suppression has been taken into account .
two year ago , beneke , buchalla , neubert , and sachrajda ( bbns ) gave a qcd factorization formula in the heavy quark limit for the decays @xmath12@xcite .
they pointed out that the radiative corrections from hard gluon exchange can be calculated by use of the perturbative qcd method if one neglects the power contributions of @xmath13 .
this factorization formula can be justified in case that the ejected meson from the @xmath8 quark decay is a light meson or an onium , no matter whether the other recoiling meson which absorbs the spectator quark in @xmath0 meson is light or heavy .
but for the case that the ejected meson is in an extremely asymmetric configuration , such as d meson , this factorization formula does not hold .
the contributions from the hard scattering with the spectator quark in b meson are also involved in their formula .
this kind of contribution can not be contained in the naive factorization .
but it appears in the order of @xmath1 .
so they said that the naive factorization can be recovered if one neglects the radiative corrections and power @xmath13 suppressed contributions in the qcd factorization , and the ` non - factorizable ' contributions in the naive factorization can be calculated perturbatively , then we do not need a phenomenological parameter @xmath14 to compensate the ` non - factorizable ' effects any more@xcite .
this qcd factorization ( bbns approach ) has been applied to study many @xmath0 meson decay modes , such as @xmath15@xcite , @xmath16 , @xmath11@xcite and other interesting channels@xcite
. some theoretical generalizations of bbns approach have also been made , such as the chirally enhanced power corrections@xcite from the twist-3 light - cone distribution amplitudes of the light pseudoscalar mesons . in this work
, we will take a closer look at this issue .
this work is organized as follows : sect .
ii is devoted to a sketch of the low energy effective hamiltonian ; in sect.iii , we will give a detailed overview of qcd factorization , in which some elaborate calculations are shown , especially for the chirally enhanced power corrections ; sect .
iv is for some detailed discussions and comparison of bbns approach to the generalized factorization and pqcd method ; we conclude in sect.v with a summary .
@xmath0 decays involve three characteristic scales which are strongly ordered : @xmath17 .
how to separate or factorize these three scales is the most essential question in b hadronic decays . with the operator product expansion method ( ope ) ,
the relevant @xmath18 effective hamiltonian is given by @xcite : @xmath19+h.c .
, \end{aligned}\ ] ] where @xmath20(for @xmath21 transition ) or @xmath22(for @xmath23 transition ) and @xmath24 are the wilson coefficients which have been evaluated to next - to - leading order approximation with the perturbative theory and renormalization group method . in the eq.(1 ) , the four - quark operators @xmath25 are given by @xmath26 and @xmath27 with @xmath28 and @xmath29 being the tree operators , @xmath30 the qcd penguin operators , @xmath31 the electroweak penguin operators , and @xmath32 , @xmath33 the magnetic - penguin operators . in this effective hamiltonian for @xmath0 decays ,
the contributions from large virtual momenta of the loop corrections from scale @xmath34 to @xmath35 are attributed to the wilson coefficients , and the low energy contributions are fully incorporated into the matrix elements of the operators@xcite .
so the derivation of the effective hamiltonian can be called `` the first step factorization '' . to evaluate the wilson coefficients
, we must extract them at a large renormalization scale ( for example @xmath36 in the standard model ) by matching the amplitude of the effective hamiltonian ( @xmath37 ) to that of the full theory ( @xmath38 ) , then evolve them by the renormalization group equations from the scale @xmath36 to the scale @xmath34 .
it should be noted that the extraction of the wilson coefficients @xmath39 by matching does not depend on the choice of the external states , if we regularize the infrared ( and mass ) singularities properly@xcite .
all dependence on the choice of external states only appears in the matrix elements @xmath40 , and is not contained in @xmath39 .
so @xmath39 only contains the short - distance contributions from the region where the perturbative theory can be applied .
but for the matrix elements @xmath40 , the long - distance contributions appear , and are process - dependent .
several years ago , the perturbative corrections to the wilson coefficients in sm have been evaluated to next - to - leading order with renormalization group method@xcite .
as we know , the wilson coefficients are generally renormalization scheme and scale dependent .
so , in order to cancel such dependence , we must calculate the hadronic matrix elements of the effective operators to the corresponding perturbative order with the same renormalization scheme and at the same scale , then we can obtain a complete decay amplitude which is free from those unphysical dependences .
after `` the first step factorization '' , the decay amplitude for @xmath42 can be written as @xmath43 in which , as mentioned in the previous section , the contributions from the large scale @xmath35 down to @xmath44 has been separated into the wilson coefficients @xmath24 .
the remaining task is to calculate the hadronic matrix elements of the effective operators .
but for the complexity of qcd dynamics , it is difficult to calculate these matrix elements reliably from first principles .
the most popular approximation is factorization hypothesis , in which the matrix element of the current - current operator is approximated to a product of two matrix elements of single current operator : @xmath45 obviously , under this approximation , the original hadronic matrix element @xmath46 misses the dependence of the renormalization scheme and scale which should be used to cancel the corresponding dependence in the wilson coefficients @xmath24 . a plausible solution to recover this scale and scheme dependence of @xmath40 is to calculate the radiative corrections . in one - loop level
, they can be written as @xcite : @xmath47\cdot \langle { \bf q } \rangle_{\rm tree } .\ ] ] here @xmath48 and @xmath49 represent the one loop corrections of qcd and qed respectively .
then one takes @xmath50 therefore , the scheme and scale dependence of @xmath51 which are expressed in the form of @xmath48 and @xmath52 is recovered .
but in quark level , @xmath48 and @xmath52 usually contain infrared divergences if we take the external quarks on - shell@xcite .
to remove or regularize the infrared divergence , the conventional treatment is to assume that external quarks are off - shell by @xmath53 . but
this introduction of the infrared cutoff @xmath53 results in a gauge dependence of one - loop corrections .
so how to factorize the infrared part of the matrix elements is a very subtle question .
but maybe this question would get an important simplification in the case that the final states of b meson decay are two light mesons .
two years ago , beneke , buchalla , neubert and sachrajda proposed a promising qcd factorization method for @xmath54 .
they pointed out that in the heavy quark limit @xmath55 , the hadronic matrix elements for @xmath12 can be written in the form @xmath56.\ ] ] obviously , the above formula reduces to the naive factorization if we neglect the power corrections in @xmath13 and the radiative corrections in @xmath1 .
they find that the radiative corrections , which are dominated by hard gluon exchange , can be calculated systematically with the perturbative theory in the limit @xmath57 , in terms of the convolution of the hard scattering kernel and the light - cone distribution amplitudes of the mesons .
this is also similar to the framework of perturbative factorization for exclusive processes in qcd at large momentum transfer , such as the calculation of the electromagnetic form factor of the pion @xcite .
then a factorization formula for @xmath12 can be written as @xcite : @xmath58 we call this factorization formalism as qcd factorization or the bbns approach . in the above formula ,
@xmath59 and @xmath60 are the leading - twist wave functions of @xmath0 and pion mesons respectively , and the @xmath61 denote hard - scattering kernels which are calculable in perturbative theory . at the order of @xmath1 , the hard kernels @xmath62 can be depicted by fig.1 .
figures 1(a)-1(d ) represent vertex corrections , figs 1(e ) and 1(f ) penguin corrections , and figs 1(g ) and 1(h ) hard spectator scattering . in the heavy quark limit ,
both pions are energetic .
the pion ejected from @xmath8 quark decay moves so fast that it can be described by its leading - twist light - cone distribution amplitude .
the @xmath63 pair in the ejected pion is produced as a small - size color dipole .
consequently , the ejected pion decouples from the soft gluons at leading order of @xmath13 .
of course , only the cancellation of soft gluons is not enough to make the factorization hold , it is necessary that the @xmath63 pair also decouples from the collinear gluons .
both the cancellations of soft gluons and collinear gluons guarantee that the hard kernel @xmath64 is of infrared finiteness .
contrast to the pion ejected from b quark weak decay , the recoiling pion which picks up the spectator in b meson can not be described by its leading - twist light - cone distribution amplitude(lcda ) , because the spectator is transferred to the recoiling pion as a soft quark . here
_ take the point of view that the form factor @xmath65 can not be calculated perturbatively .
if we attempt to calculate the form factor within the perturbative framework , by the naive power counting , we find that the leading twist lcda of pion does not fall fast enough to suppress the singularity at the endpoint where the quark from @xmath8 decay carries almost all momentum of the pion .
it indicates that the contributions to form factor are dominated by the soft gluon exchange@xcite .
this point of view can be justified also from the calculation of the form factor @xmath66 by using light - cone sum rule ( lcsr)@xcite , in which the dominated contribution to @xmath65 comes from the region where the the spectator quark is transferred as a soft quark to the pion .
so the transition form factor survives in the factorization formula as a nonperturbative parameter
. however , when the spectator quark in @xmath0 meson interacts with a hard gluon from the ejected pion , the recoiling pion can be also described by its light - cone distribution amplitude .
this hard spectator scattering is missed in the naive factorization , but calculable in the perturbative qcd at the leading power in @xmath13 .
so with this factorization formula , the remaining hard part of the hadronic matrix element @xmath40 from the scale about @xmath44 has been factorized into the hard scattering kernel , and the long distance contributions are absorbed into the transition form factors and the light - cone wave functions of the participating mesons .
thus this is the `` final factorization '' for the two - body nonleptonic charmless @xmath0 decays .
an explicitly technical demonstration of the above argument has been presented in one - loop level in refs.@xcite . for @xmath67 ,
this qcd factorization has been proved to two - loop order@xcite . in the literature ,
the ejected pion is represented by its leading twist light - cone distribution amplitude(lcda ) . however , since the mass of @xmath8 quark is not asymptotically large , in particular , some power corrections might be enhanced by certain factors , such as the scale of chiral symmetry breaking @xmath68 gev , and have significant effects in studying @xmath0 two - body nonleptonic charmless decays .
so , in this manner , the chirally enhanced power corrections must be taken into account .
accordingly , describing the ejected pion by its leading twist lcda is not enough , the two - particle twist-3 lcdas must be taken into account .
below , we will show the elaborate results of qcd factorization in these two cases . for illustration
, we take @xmath69 as an example , but the result is easily generalized to the cases that the final states are the other light pseudoscalars .
when inserting leading - twist lcda of the light pseudoscalar , in the heavy quark limit , the quark constituents of the ejected pion can be treated as a pair of collinear massless quark and antiquark with the momentum @xmath70 and @xmath71 respectively ( @xmath72 is the momentum of the ejected pion and we take @xmath72 as a hard light - cone momentum in calculation , @xmath73 ) , because that the contributions from the transverse momenta of the quarks in ejected pion are power suppressed @xcite .
now we move on to the explicit one - loop calculation of the diagram figs .
1(a)-1(d ) for @xmath74 . for illustration
, we write down the one - gluon exchange contribution to the @xmath75 matrix element of the operator @xmath76 . @xmath77 } \nonumber \\ & & \times \langle \pi^+ \vert \bar{u}_i \gamma^{\mu } ( 1-\gamma_5)\slash{\hskip -2 mm } q \gamma_5 \gamma^{\alpha } ( u\slash{\hskip -2 mm } q-\slash{\hskip -2.5 mm } k)\gamma_{\mu}(1-\gamma_5 ) ( \slash{\hskip -2.5 mm } p-\slash{\hskip -2.5mm}k+m_b ) \gamma_{\alpha } b_i \vert \bar{b}_d^0\rangle , \\
\langle q_2^u\rangle_{(b)}&=&g_s^2\frac{f_{\pi}}{4 } \frac{c_f}{n } \int_0 ^ 1 du \phi(u ) \int \frac{d^4k}{(2\pi)^4 } \frac{1}{k^2(\bar{u}q - k)^2[(p - k)^2-m_b^2 ] } \nonumber \\ & & \times \langle \pi^+ \vert\bar{u}_i \gamma^{\mu } ( 1-\gamma_5)(\bar{u}\slash{\hskip -2 mm } q- \slash{\hskip -2.5mm}k)\gamma^{\alpha}\slash{\hskip -2 mm } q \gamma_5
\gamma_{\mu}(1-\gamma_5 ) ( \slash{\hskip -2.5 mm } p-\slash{\hskip -2.5mm}k+m_b ) \gamma_{\alpha } b_i \vert \bar{b}_d^0\rangle , \\
\langle q_2^u \rangle_{(c)}&=&-g_s^2\frac{f_{\pi}}{4 } \frac{c_f}{n } \int_0 ^ 1 du \phi(u ) \int \frac{d^4k}{(2\pi)^4 } \frac{1}{k^2(uq+k)^2(p - q - k)^2 } \nonumber \\ & & \times \langle \pi^+ \vert\bar{u}_i \gamma_{\alpha } ( \slash{\hskip -2.5 mm } p- \slash{\hskip -2 mm } q-\slash{\hskip -2.5 mm } k ) \gamma^{\mu } ( 1-\gamma_5)\slash{\hskip -2 mm } q \gamma_5 \gamma_{\alpha } ( u \slash{\hskip -2 mm } q + \slash{\hskip -2.5mm}k ) \gamma_{\mu } ( 1-\gamma_5 ) b_i \vert \bar{b}_d^0\rangle , \\
\langle q_2^u \rangle_{(d)}&=&g_s^2\frac{f_{\pi}}{4 } \frac{c_f}{n } \int_0 ^ 1 du \phi(u ) \int \frac{d^4k}{(2\pi)^4 } \frac{1}{k^2(\bar{u}q+k)^2(p - q - k)^2 } \nonumber \\ & & \times \langle \pi^+ \vert\bar{u}_i \gamma_{\alpha}(\slash{\hskip -2.5 mm } p- \slash{\hskip -2 mm } q-\slash{\hskip -2.5 mm } k ) \gamma^{\mu } ( 1-\gamma_5 ) ( \bar{u}\slash{\hskip -2mm}q+ \slash{\hskip -2.5mm}k ) \gamma_{\alpha } \slash{\hskip -2 mm } q \gamma_5 \gamma_{\mu } ( 1-\gamma_5 ) b_i \vert \bar{b}_d^0\rangle , \end{aligned}\ ] ]
when we calculate the vertex corrections in the leading power of @xmath13 , not only ultraviolet divergence emerges but infrared divergence does also
. infrared divergence arises from two regions where the virtuality of the loop @xmath78 is soft or collinear to the momentum of the pions . in ref.@xcite ,
the authors gave an explicit cancellation of soft and collinear divergence in vertex corrections for @xmath79 in eikonal approximation .
figures 1(a),1(b ) and 1(c),1(d ) cancel the soft divergence ; 1(a),1(c ) and 1(b),1(d ) cancel the collinear divergence . for @xmath74 ,
the cancellation is similar except that the collinear divergence also arises from the region where @xmath78 is collinear to the momentum of the recoiling pion .
1(c),1(d ) cancel not only part of soft divergence but also part of collinear divergence .
below , we give an explicit calculation of the feynman diagrams figs .
1(a)-1(d ) to show the cancellation of the infrared divergences .
in order to regularize the infrared divergence , there are two choices for us .
one is the dimensional regularization ( dr ) scheme , in which the infrared divergence can be regularized into the pole terms @xmath80 .
in contrast to the dimensional regularization of ultraviolet divergence , the infrared divergence arises when @xmath81 , instead of @xmath82 in the case of the ultraviolet divergence . so the dimension @xmath83 in regularization for infrared divergence must be set to be greater than 4 .
this is a subtle point , but it will not cause any ambiguity in our calculation because the infrared part and ultraviolet part can be safely separated .
the other method to regularize the infrared divergence is the well - known massive gluon ( mg ) scheme , in which the infrared divergence is handled by replacing @xmath84 by @xmath85 in the gluon propagator .
similar scheme has been applied in earlier computation of the radiative corrections for @xmath86 , in which the massless photon is replaced by a massive photon . in addition , in our latter calculation , there are also several schemes in treating @xmath87 , the most popular two are the naive dimensional regularization ( ndr ) scheme and the t hooft - veltman renormalization ( hv ) scheme .
both have been applied to calculate the wilson coefficients @xcite . in this work ,
if there is no specification , the ndr scheme is always applied in our calculations for its simplicity . after a straightforward calculation in dr scheme and using the corresponding feynman parameter integrals listed in appendix c , we obtain @xmath88 \right .
\nonumber \\ & - & \left .
\frac{\gamma(1-a)}{(4 \pi)^a } \left(\frac{m_b}{\mu}\right)^{2a } \left [ \frac{1}{a^2}+\frac{2(\ln u -1)}{a}+\ln^2 u- 2{\rm li}_2(1-\frac{1}{u } ) - 4\ln u+5+\frac{2\ln u}{1-u}\right ] \right\ } , \\
q_2^u \rangle_{(b)}&=&\frac{\alpha_s}{4\pi}\frac{c_f}{n}\langle \pi^- \vert \bar{d}_{\alpha}\gamma^{\mu}(1-\gamma_5)u_{\alpha}\vert 0\rangle \langle \pi^+ \vert \bar{u}_{\beta}\gamma_{\mu}(1-\gamma_5)b_{\beta}\vert \bar{b}_d^0 \rangle \nonumber \\ & \times & \int_0 ^ 1 du \phi(u ) \left\ { -4 \left[\frac{1}{\epsilon}-\gamma_e+\ln 4\pi+2\ln \frac{\mu}{m_b}+ \frac{11}{4}+ \frac{\bar{u}}{1-\bar{u } } \ln \bar{u}\right ] \right .
\nonumber \\ & + & \left .
\frac{\gamma(1-a)}{(4 \pi)^a } \left(\frac{m_b}{\mu}\right)^{2a } \left [ \frac{1}{a^2}+\frac{2(\ln \bar{u } -1)}{a}+\ln^2 \bar{u}- 2{\rm li}_2(1-\frac{1}{\bar{u } } ) - 4\ln\bar{u}+6+\frac{2\ln \bar{u}}{1-\bar{u}}\right ] \right\},\\ \langle q_2^u \rangle_{(c)}&=&\frac{\alpha_s}{4\pi}\frac{c_f}{n}\langle \pi^- \vert \bar{d}_{\alpha}\gamma^{\mu}(1-\gamma_5)u_{\alpha}\vert 0\rangle \langle \pi^+ \vert \bar{u}_{\beta}\gamma_{\mu}(1-\gamma_5)b_{\beta}\vert \bar{b}_d^0
\rangle \nonumber \\ & \times & \int_0 ^ 1 du \phi(u ) \left\ { -4\left[\frac{1}{\epsilon}-\gamma_e+\ln 4\pi+2\ln \frac{\mu}{m_b}+ \frac{11}{4}-\ln(-u)\right ] \right .
\nonumber \\ & + & \left .
\frac{\gamma(1-a)}{(4 \pi)^a } \left(\frac{m_b}{\mu}\right)^{2a } \left [ \frac{2}{a^2}+\frac{2(\ln ( -u ) -2)}{a}+ 10-\frac{\pi^2}{3}-4\ln(-u)+\ln^2(-u)\right ] \right\ } , \\
\langle q_2^u \rangle_{(d)}&=&\frac{\alpha_s}{4\pi}\frac{c_f}{n}\langle \pi^- \vert \bar{d}_{\alpha}\gamma^{\mu}(1-\gamma_5)u_{\alpha}\vert 0\rangle \langle \pi^+ \vert \bar{u}_{\beta}\gamma_{\mu}(1-\gamma_5)b_{\beta}\vert \bar{b}_d^0 \rangle \nonumber \\ & \times & \int_0 ^ 1 du \phi(u ) \left\ { \left[\frac{1}{\epsilon}-\gamma_e+\ln 4\pi+2\ln \frac{\mu}{m_b}+ 1-\ln(-\bar{u})\right ] \right . \nonumber \\ & - & \left .
\frac{\gamma(1-a)}{(4 \pi)^a } \left(\frac{m_b}{\mu}\right)^{2a } \left [ \frac{2}{a^2}+\frac{2(\ln ( -\bar{u } ) -2)}{a}+ 10-\frac{\pi^2}{3}-4\ln(-\bar{u})+\ln^2(-\bar{u})\right ] \right\}. \end{aligned}\ ] ] in above , we have set @xmath89 ( @xmath90 ) in regularizing the infrared divergence .
then , after summing over all four diagrams , we find that all pole terms in @xmath91 are really cancelled before we integrate over the momentum fraction variable @xmath92 .
so after modified minimal subtraction ( @xmath93 ) , we get @xmath94 . \label{lwas } \end{aligned}\ ] ] assuming that the light - cone distribution amplitude @xmath95 is symmetric , then the above equation can be simplified as follows : @xmath96.\label{lws } \end{aligned}\ ] ] it is easy to check that the above equations are consistent with the results in previous works . actually ,
with the mg scheme , we get the same results as that by using the dr scheme . with eqs.([lwas],[lws ] ) , we can compute the vertex corrections no matter the lcda @xmath95 is symmetric or asymmetric .
this is very important in principle .
for instance , when kaon is ejected from b quark decay , we must take the contributions from the asymmetric part of lcda of kaon into account , although the contributions from the asymmetric part are very small numerically@xcite .
there are two kinds of penguin corrections .
one is the four quark operators insertion [ fig .
1(e ) ] ; the other is the magnetic penguin insertion [ fig .
1(f ) ] . the first kind is generally called bss mechanism . in generalized factorization , bss mechanism plays a very important role in @xmath3 violation because it is the unique source of strong phases .
but in generalized factorization , the virtuality of the gluon or photon is ambiguous ; usually one varies @xmath97 around @xmath98 .
this variation does not have an important effect on the branching ratios , but it does for @xmath3 asymmetries . in qcd factorization , this ambiguity is rendered by taking the virtuality of the gluon as @xmath99 . when treating penguin contractions , one should be careful that fig .
1(e ) contains two kinds of topology , which is depicted in fig .
they are equivalent in 4 dimensions according to fierz rearrangement . however , since penguin corrections contain ultraviolet divergence , we must do calculations in d dimensions where these two kinds of topology are not equivalent @xcite .
below we give an explicit calculation of @xmath100 or @xmath101 penguin insertions for @xmath102 which belong to the second topology , fig .
2(b ) : @xmath103}{[(l - k)^2-m_q^2][k^2-m_q^2]l^2 } \right |_{l = p - uq } \nonumber \\ & = & -2i f_{\pi}\frac{\alpha_s}{4 \pi } \frac{c_f}{n } \int_0 ^ 1 du \phi(u ) \langle \pi^+\vert \bar{u}_i \gamma_{\alpha } \slash{\hskip -2.0mm}q \gamma_5 \gamma_{\mu } ( 1- \gamma_5 ) b_i \vert \bar{b}_d^0 \rangle \left[\frac{l_{\alpha}l_{\mu}}{l^2}-g_{\alpha \mu}\right ] \nonumber \\ & \times & \left .
\sum \limits_q \left [ \frac{1}{6}(\frac{1}{\epsilon}-\gamma_e+\ln 4 \pi ) + \int_0 ^ 1 dt~t(1-t)\ln\frac{\mu^2}{m_q^2-t(1-t)l^2-i \epsilon}\right]\right|_{l = p - uq}~. \end{aligned}\ ] ] after @xmath93 subtraction and using the equations of motions , we get the finite result @xmath104~,\end{aligned}\ ] ] where @xmath105 .
the first topology , fig .
2(a ) , for example , @xmath106 penguin insertion for @xmath107 , is similar to the results of the second topology , fig.2(b ) , except that there is an extra factor @xmath108 : @xmath109~.\end{aligned}\ ] ] for the magnetic penguin insertion , it is the easiest calculation of the radiative corrections .
the result of @xmath33 insertion for @xmath102 is @xmath110 hard spectator scattering [ fig.1(g ) and ( h ) ] is completely missing in the naive factorization .
but in qcd factorization , it can be calculated in the perturbative qcd , and expressed by a convolution of the hard kernel @xmath111 and the lcdas of mesons . at the leading power of @xmath13 ,
both of the light pseudoscalars from the @xmath0 meson decay can be represented by their leading twist lcdas .
so after a straightforward calculation , we obtain this contribution for @xmath102 from the operator @xmath112 insertion , @xmath113}{k^2 l_d^2 } \right |^{k=\xi p-\bar{v}q_1}_{l_d = uq_2-k } \right .
\nonumber \\ & & + \left . \left .
\frac{{\rm tr } [ ( \sl p + m_b)\gamma_5 \gamma^{\alpha } \sla q_1 \gamma_5 \gamma_{\rho } ( 1-\gamma_5 ) \slash{\hskip -1.5 mm } l_u \gamma_{\alpha } \sla q_2 \gamma_5 \gamma_{\rho}(1-\gamma_5)]}{k^2 l_u^2 } \right |^{k=\xi p-\bar{v}q_1}_{l_u = k-\bar{u}q_2 } \right\ } \nonumber \\ & = & i \pi \alpha_s f_b f_{\pi}^2 \frac{c_f}{n^2 } \int_0 ^ 1 d\xi\frac{\phi_b(\xi)}{\xi}\int_0 ^ 1 ~du \frac{\phi(u)}{u } \int_0 ^ 1~dv ~ \frac{\phi(v)}{\bar{v } } ~.\end{aligned}\ ] ] it has been observed that qcd factorization is demonstrated only in the strict heavy quark limit .
this means that any generalization of qcd factorization to include or partly include power corrections of @xmath13 should redemonstrate that factorization still holds .
there are a variety of sources which may contribute to power corrections in @xmath114 ; examples are higher twist distribution amplitudes , transverse momenta of quarks in the light meson , annihilation diagrams , etc .
unfortunately , there is no known systematic way to evaluate these power corrections for exclusive decays . moreover
, factorization might break down when these power corrections , for instance , transverse momenta effects , are considered .
this indicates that one might have to give up such an ambitious plan that all power corrections could be , at least in principle , incorporated into qcd factorization order by order .
one might argue that power corrections in @xmath0 decays are numerically unimportant because these corrections are expanded in order of a small number @xmath115 .
but this is not true .
for instance , the contributions of operator @xmath101 to decay amplitudes would formally vanish in the strict heavy quark limit .
however it is numerically very important in penguin - dominated @xmath0 rare decays , such as interesting channels @xmath116 , etc .
this is because @xmath101 is always multiplied by a formally power suppressed but chirally enhanced factor @xmath117 , where @xmath118 and @xmath119 are current quark masses .
so power suppression might probably fail at least in this case
. therefore phenomenological applicability of qcd factorization in @xmath0 rare decays requires at least a consistent inclusion of chirally enhanced corrections .
chirally enhanced corrections arise from the two particle twist-3 light - cone distribution amplitudes , generally called @xmath120 and @xmath121 .
so when chirally enhanced corrections are concerned , the final light mesons should be described by leading twist and twist-3 distribution amplitudes . then it is crucial to show that factorization really holds when considering twist-3 distribution amplitudes .
the most difficult part is to demonstrate the infrared finiteness of the hard scattering kernels @xmath64 .
in addition , possible chirally enhanced power corrections can also appear in the hard spectator scattering .
so , for consistency , we must involve these corrections .
when we calculate the chirally enhanced power corrections , contrast to the leading - twist light - cone wave function insertion , the cancellation of the infrared divergences in the vertex corrections to @xmath122 operator ( here it is @xmath123 or @xmath124 ) can not be shown simply by the eikonal approximation similar to what has been done at the leading power of @xmath13 , because the dirac structure or spin structure of twist-3 light - cone wave functions of the light pseudoscalar makes the `` on - shell '' condition for the external quarks invalid .
thus , to justify the cancellation of the infrared divergence in @xmath122 vertex corrections , we must give the explicit calculation .
as mentioned in the previous subsections , we have two choices to regularize the infrared divergence in one - loop calculation .
one is the dr scheme ; the other is mg scheme .
generally , these two schemes are equivalent , for instance , similar to what has been done in @xmath125 vertex corrections . however , in the dr scheme , it is difficult to extrapolate the twist-3 wave functions of the light pseudoscalar to d dimensions properly , although they are well - defined in 4 dimensions .
therefore , we prefer to use the mg scheme in our calculation for chirally enhanced corrections to avoid the above possible problems .
in addition , generally we calculate the feynman diagrams in the momentum space , so the correct projection of the light - cone wave functions of the light pseudoscalar in the momentum space is necessary . from appendix b
, we find that it is easy to find the proper momentum space projection of the leading twist and @xmath126 type twist-3 wave function , but for @xmath127 , the projection is not very clear .
note that the coordinate @xmath128 in the definition of @xmath127 by the non - local matrix element must be transformed into a partial derivative of a certain momentum in the projection of momentum space .
however , it is difficult to find the derivative which makes the projection only depend on the structure of the light pseudoscalar itself . generally , the momentum which the partial derivative acts on is dependent on the hard kernel .
therefore , we prefer to compute the feynman diagrams of the twist-3 wave functions insertion , especially @xmath127 insertion in the coordinate space .
we think that such calculation can avoid the ambiguity about how to project the coordinate @xmath128 into the momentum space .
we recalculate the leading twist insertion by using the same method , and obtain the same results as those in the previous sections .
below , we will show how to perform this trick in calculation of @xmath127 insertion .
for example , let us consider fig .
3 . in coordinate space
, we have @xmath129 the above trick has been applied in the calculation of the proper correlation function to extract the transition form factor @xmath65 within the frame of the light - cone sum rule@xcite .
so within the same frame , we obtain @xmath130 perform the one - loop integrations : @xmath131\right .
\nonumber \\ & + & \left .
\frac{\phi_{\sigma}(u)}{6u } \left [ \frac{1}{2}\ln^2 \lambda + 2 \ln(-u)\ln\lambda -4 \ln u \ln \lambda+\ln \lambda \right .
\nonumber \\ & + & \left .
\left.4\ln^2 u-\ln^2(-u)-2\ln u \ln(1-u)+2{\rm li}_2(\frac{1}{u } ) + \frac{7}{6}\pi^2-\ln u -\frac{\ln u}{1-u}\right]\right\ } , \\
\langle q_5 \rangle_{(b)}&=&-2\frac{\alpha_s}{4 \pi } \frac{c_f}{n } \langle\pi^- \vert \bar{d}_i ( 1+\gamma_5 ) u_i\vert 0\rangle \langle \pi^+ \vert \bar{u}_j ( 1-\gamma_5 ) b_j \vert \bar{b}_d^0 \rangle \nonumber \\ & \times&\int_0 ^ 1 du~ \left\{\phi_p(u)\left[-(\frac{1}{\epsilon}-\gamma_e+\ln 4 \pi + 2\ln \frac{\mu}{m_b } ) \right.\right.\nonumber \\ & + & \left .
\left.\frac{1}{4}\ln^2 \lambda+\ln(-\bar{u})\ln\lambda -2\ln \bar{u } \ln\lambda + \frac{1}{2}\ln^2 \bar{u}-{\rm li}_2(1-\frac{1}{\bar{u } } ) -\frac{5}{2}+\frac{5}{4 } \pi^2\right]\right .
\nonumber \\ & + & \left.\frac{\phi_{\sigma}(u)}{6\bar{u } } \left [ \frac{1}{2}\ln^2 \lambda + 2 \ln(-\bar{u})\ln\lambda- 4 \ln \bar{u } \ln \lambda + \ln \lambda \right . \right .
\nonumber \\ & + & \left .
4\ln^2 \bar{u}-\ln^2(-\bar{u})-2\ln \bar{u } \ln(1-\bar{u})+2{\rm li}_2(\frac{1}{\bar{u } } ) + \frac{7}{6}\pi^2-\ln \bar{u } -\frac{\ln \bar{u}}{1-\bar{u}}\right]\right\ } , \\
\langle q_5\rangle_{(c)}&=&-2\frac{\alpha_s}{4 \pi } \frac{c_f}{n } \langle\pi^- \vert \bar{d}_i ( 1+\gamma_5 ) u_i\vert 0\rangle \langle \pi^+ \vert \bar{u}_j ( 1-\gamma_5 ) b_j \vert \bar{b}_d^0 \rangle \nonumber \\ & \times&\int_0 ^ 1 du~ \left\{\phi_p(u)\left[-(\frac{1}{\epsilon}-\gamma_e+\ln 4\pi+2\ln\frac{\mu}{m_b } ) \right . \right .
\nonumber \\ & + & \left .
\left.\frac{1}{2}\ln^2\lambda-[\ln(-u)-2]\ln \lambda+ \frac{1}{2}\ln^2(-u)-\ln(-u)-\frac{3}{2}+\frac{\pi^2}{3}\right]\right .
\nonumber \\ & + & \left .
\frac{\phi_{\sigma}(u)}{6u } \left [ \ln^2 \lambda - [ 2\ln(-u)-3]\ln\lambda \right .
\right.\nonumber \\ & + & \left .
2\ln(-u)\ln u -\ln^2 u-3\ln(-u ) + 3-\frac{\pi^2}{3 } \right ] \right\ } , \\
\langle q_5 \rangle_{(d)}&=&2\frac{\alpha_s}{4 \pi } \frac{c_f}{n } \langle\pi^- \vert \bar{d}_i ( 1+\gamma_5 ) u_i\vert 0\rangle \langle \pi^+ \vert \bar{u}_j ( 1-\gamma_5 ) b_j \vert \bar{b}_d^0 \rangle \nonumber \\ & \times&\int_0 ^ 1 du~ \left\{\phi_p(u)\left[-(\frac{1}{\epsilon}-\gamma_e+\ln 4\pi+2\ln\frac{\mu}{m_b } ) \right . \right .
\nonumber \\ & + & \left .
\left.\frac{1}{2}\ln^2\lambda-[\ln(-\bar{u})-2]\ln \lambda+ \frac{1}{2}\ln^2(-\bar{u})-\ln(-\bar{u})+\frac{3}{2}+\frac{\pi^2}{3}\right]\right .
\nonumber \\ & + & \left .
\frac{\phi_{\sigma}(u)}{6\bar{u } } \left [ \ln^2 \lambda - [ 2\ln(-\bar{u})-3]\ln\lambda \right .
\right.\nonumber \\ & + & \left .
2\ln(-\bar{u})\ln \bar{u } -\ln^2 \bar{u}-3\ln(-\bar{u } ) + 3-\frac{\pi^2}{3 } \right ] \right\}. \end{aligned}\ ] ] here @xmath132 . from the above equations , it is observed that , in the case of @xmath127 distribution amplitudes , the terms with infrared divergence in vertex correction diagrams can not cancel unless @xmath133 is a symmetric function : @xmath134
this is an unexpected result , which means qcd factorization is violated for asymmetric twist-3 light - cone distribution amplitudes .
this indicates that chirally enhanced corrections can be included consistently in the framework of qcd factorization only when twist-3 light - cone distribution amplitudes are symmetric .
therefore , in the following , we will implicitly assume a symmetric twist-3 light - cone distribution amplitude for light pseudoscalar mesons .
it is then straightforward to show that vertex corrections of @xmath122 operator are completely cancelled after summing over four diagrams in the case of @xmath127 distribution amplitude .
the final result of @xmath122 vertex corrections , in the condition that the twist-3 lcda is symmetric , is @xmath135 in quark level , usually one decomposes the basic qcd vertex @xmath136 in penguin insertion into the two chiral current couplings @xmath137 and @xmath138 ; then the penguin insertions contribute the same magnitude to the @xmath125 and @xmath122 vertex .
but in hadron level , this point of view must be examined in elaborate calculation . for illustration
, we give the results of @xmath100 or @xmath101 penguin corrections to @xmath102 , which belong to the second penguin topology fig .
2(b ) , when @xmath139 is inserted : @xmath140}{[(p - uq - k)^2-m_q^2][k^2-m_q^2](p - uq)^2 } \nonumber \\ & = & 2i f_{\pi}\mu_{\pi}\frac{\alpha_s}{4 \pi } \frac{c_f}{n } \int_0 ^ 1 du~ \phi_p(u ) \langle \pi^+\vert \bar{u}_i \gamma_{\alpha } \gamma_5 \gamma_{\mu } ( 1- \gamma_5 ) b_i \vert \bar{b}_d^0 \rangle \left[\frac{l_{\alpha}l_{\mu}}{l^2}-g_{\alpha \mu}\right ] \nonumber \\ & \times & \left .
\sum \limits_q \left [ \frac{1}{6}(\frac{1}{\epsilon}-\gamma_e+\ln 4 \pi ) + \int_0 ^ 1 dt~t(1-t)\ln\frac{\mu^2}{m_q^2-t(1-t)l^2-i\epsilon } \right]\right|_{l = p - uq } \nonumber \\ \end{aligned}\ ] ] after @xmath93 subtraction , we obtain @xmath141~.\end{aligned}\ ] ] for the first penguin insertion topology , fig .
2(a ) , the result is @xmath142~.\end{aligned}\ ] ] similarly , when @xmath133 is inserted , by using the method in the previous subsection , we have @xmath143}{[(l_q - k)^2-m_q^2][l_q^2-m_q^2]k^2}\right ] \right|_{k = p - uq } .
\end{aligned}\ ] ] after integration and subtraction , @xmath144 .
\end{aligned}\ ] ] the magnetic penguin insertion is easier ; we write the result of @xmath33 insertion as follows : @xmath145~[\gamma^{\sigma},\gamma_{\alpha}](1+\gamma_5)b_i \vert \bar{b}_d^0 \rangle \right .
\right .
\nonumber \\ & -&\left.\left.\left .
\frac{2}{k^2 } \langle \pi^+\vert \bar{u}_i \gamma^{\alpha}\gamma_5 [ \sl q , \sl k]~[\sl k,\gamma_{\alpha}](1+\gamma_5)b_i \vert \bar{b}_d^0 \rangle \right ] \right\}\right |_{k = p - uq } \nonumber \\ & = & 2\frac{\alpha_s}{4\pi}\frac{c_f}{n}(\frac{3}{2}+\int_0 ^ 1 du~\frac{\phi_{\sigma}(u)}{6\bar{u } } ) \langle \pi^- \vert \bar{d}_i ( 1+\gamma_5 ) u_i\vert 0 \rangle \langle \pi^+ \vert \bar{u}_j ( 1-\gamma_5 ) b_j \vert \bar{b}_d^0 \rangle .
\end{aligned}\ ] ] the chirally enhanced power corrections in hard spectator scattering not only occurs in the case of @xmath122 vertex insertion , but also in the case of @xmath125 insertion .
but in the case of @xmath122 insertion , after a straightforward calculation , we find that there will be serious linear divergence at the end points of the lcdas if the twist-3 lcdas are not symmetric .
because infrared finiteness of the vertex corrections requires that the twist-3 lcdas , especially @xmath133 , must be symmetric , we shall implicitly assume this symmetric condition for the lcdas in latter computation .
so , in this symmetric condition , the hard scattering with the spectator vanishes when @xmath122 vertex is inserted .
however , even in this strict symmetric condition , there is still a logarithmic divergence from the end point of the recoiling pion in hard spectator scattering when @xmath125 vertex is inserted .
for example , @xmath146.\ ] ] this means that qcd factorization is broken down .
but we can still give a phenomenological treatment for this hard spectator scattering . by using the asymptotic form of @xmath133
, we find that there is a divergent integral over @xmath147 : @xmath148 . in refs.@xcite ,
the authors prefer to introduce a phenomenological parameterization for this logarithmic divergence .
they take @xmath149 , where @xmath150 is taken from 3 ( realistic ) to 6 ( conservative ) , and the phase @xmath151 from @xmath152 to @xmath153 .
we shall take similar phenomenological treatment in the numerical computation below .
we notice that the above approach of evaluating hard spectator contribution is naive .
for instance , the scale of hard spectator contribution should be different from the vertex correction contribution .
while it seems reasonable to take the scale @xmath154 for the vertex correction diagrams to avoid large logarithm @xmath155 , a natural choice of the scale of hard spectator contribution may be around @xmath156 because the average momentum squared of the exchanged gluon is about @xmath157 .
another disturbing feature of hard spectator contribution is that , as pointed out in refs .
@xcite , when including the contribution of @xmath127 , there would appear a divergent integral @xmath158 even if the symmetric distribution amplitude is applied .
this divergent integral implies that the dominant contribution comes from the end - point region , or , in other words , it is dominated by soft gluon exchange . however , the transverse momentum may not be omitted in the end - point region @xcite ; if so , the corresponding divergent integral would then be changed to @xmath159 as an illustration , we do not consider the @xmath160 dependence of wave functions ( though it is certainly not a good approximation ) ; then the above integral is proportional to @xmath161 the above integration converges now ; furthermore it is not dominated by end - point contribution .
this illustrates that the treatment of hard spectator diagrams may need further discussion .
there exists `` annihilation topology '' contributions which may belong to chirally enhanced corrections . in ref .
@xcite , the authors have discussed this topic and find that a divergent integral @xmath162 ^ 2 $ ] will appear .
we suspect that this divergence may disappear , similar to the hard spectator term , if the effect of transverse momenta can be included .
it is also possible that `` annihilation topology '' contributions are really dominated by soft interactions and thus violate factorization . due to its complexity
, we do not include `` annihilation topology '' contributions in this work . with these effective operators , @xmath163 decay amplitudes in qcd factorization
can be written as @xmath164 where @xmath165 is ckm factor , @xmath166b@xmath167 is the factorized matrix element and is the same as the definition of the bsw largrangian@xcite .
then as an illustration , the explicit expressions of @xmath168 ( @xmath169 to @xmath170 ) for @xmath171 ( using symmetric lcdas of the pion ) are obtained .
but it is easy to generalize these formulas to the case that the final states are other light pseudoscalars .
furthermore , we take only part of qed corrections into account in our final formula , in particular the qed penguin insertions . now @xmath168 for @xmath12 in ndr @xmath87 scheme is listed as follows : @xcite which is free from the gauge dependence and infrared structure of the theory .
thus , in principle , we can obtain the results in the hv scheme just by using @xmath172 . in @xcite ,
the constant matrix has been applied to obtain the results in the ndr and hv scheme for the coefficients @xmath173 which only contain the current - current vertex corrections . but whether we can obtain the expression of @xmath174 or @xmath175 in hv scheme in a similar way needs further discussion . ]
@xmath176 here @xmath177 is the number of color , @xmath178 is the factor of color , @xmath105 and we define the other symbols in the above expressions as @xmath179 where @xmath180 [ @xmath120 , @xmath121 ] is leading twist ( twist-3 ) lcda of the ejected pion , and the hard - scattering functions are @xmath182 the contributions from the hard spectator scattering [ figs .
1(g),(1)(h ) ] are reduced to the factor @xmath183 : @xmath184.\ ] ] there contains a divergent integral in @xmath183 .
here we simply assume that @xmath185 ( similar to what has been done in refs .
@xcite , though our assumption here is certainly an oversimplification ) .
we thus illustrate numerically the scale dependence of @xmath186 in table 1 . here
we use the asymptotic form of the lcdas of the light pseudoscalar meson which are listed in appendix a , and the other input parameters are taken as follows@xcite : @xmath187 , @xmath188 , @xmath189 , the pole masses @xmath190 , @xmath191 , the @xmath192 masses @xmath193 , @xmath194 , @xmath195 , @xmath196 and @xmath197 .
color transparency gives a clear physics picture of qcd factorization . in the argument of the color transparency , a fast - moving small color singlet formed by a pair of @xmath63 decouples from the surrounding soft gluons
of course , as mentioned in the previous section , only the decoupling with the soft gluons is not enough for a factorization formula , the decoupling from the collinear divergence is also necessary .
we really find that both of the requirements can be satisfied in the one - loop calculations .
so the qcd factorization is guaranteed .
therefore , the calculations in the above sections seem to be a one - loop technical manifestation ( or demonstration ) of the color transparency . on the other hand , at the leading power of @xmath13 , the soft or collinear gluons only `` see '' the direction of the light meson , but are `` blind '' to the spins of the quark constituents . so the soft or collinear gluon can not distinguish whether the ejected meson from @xmath8 quark decay is a light pseudoscalar or a light longitudinally polarized vector meson . as a consequence ,
the cancellation of the infrared divergence is universal for @xmath0 decays to two light mesons , no matter whether the meson is a pseudoscalar meson or a vector meson .
therefore , the qcd factorization formula for @xmath41 at the leading power of @xmath13 is easy to be generalized to @xmath198 and @xmath199 .
similarly , the color transparency argument can not only be applied to the strong interactions , but also be generalized to the electromagnetic interactions . when the ejected meson is electric neutral , the soft photons also decouple from the fast moving small electric dipole .
so qed vertex corrections are also of infrared finiteness .
but for the case that the ejected meson is charged , qed corrections are infrared divergent , and the infrared divergence must be cancelled by the soft photon emission mechanism , which is common in the calculation of the radiative corrections for @xmath200 .
about this , it is easy to be covered in the calculation in the previous section , just replacing the qcd vertex by a qed vertex .
this can be called a one - loop demonstration for `` charge '' transparency .
it should be noted that the above arguments must be based on the condition that the ejected meson is in a very compact configuration , then it , as a small color dipole , is disentangled with the soft gluons . otherwise , if its size is too large , it is difficult to decouple from the soft gluons .
for example , the spectator quark in d meson is very soft , and runs around c quark like a soft quark cloud , which has a large overlap with the @xmath0 meson spectator system@xcite . as a consequence ,
the process in which d meson is ejected from @xmath8 decay is dominated by the soft gluon exchange . from the expressions of the qcd coefficients
@xmath168 obtained in previous sections , it is found that the renormalization scale dependence of the hadronic matrix elements of the effective operators is recovered .
apparently , we expect this recovered dependence can cancel the scale dependence of the wilson coefficients @xmath39 . with the renormalization group equations for the wilson coefficients @xmath24 at leading order logarithm approximation@xcite , @xmath201 we do find @xmath202 when we neglect the contributions from higher order of @xmath1 . but for @xmath203 or @xmath204 , some scale dependence at the order of @xmath1 still remains .
note that other qcd coefficients ( @xmath205 ) are multiplied by the product of the matrix elements of the conserved currents which are independent of the renormalization scale ; whereas the coefficient @xmath174 or @xmath175 is multiplied by a product of the two matrix elements of scalar and pseudoscalar current @xmath206 which is still of scale dependence .
this scale dependence is generally represented by the factor @xmath207 after we apply the equations of motion to transform the @xmath208 matrix elements into the type of @xmath125 . here
@xmath118 and @xmath119 are the current masses of the valence quarks in meson @xmath209 . with the renormalization group equations for the running mass of the current quark @xmath210 we have @xmath211 consequently
, we find @xmath212 with the asymptotic form of @xmath213 .
then , as we expect , the decay amplitude for @xmath0 decays to two light pseudoscalars predicted by qcd factorization is really independent of the renormalization scale within the constraint @xmath214 .
this also can be obviously seen from the numerical results of @xmath215 listed in table 1 . in particular , if we think that the results of qcd factorization are reliable and really independent of the renormalization scale , maybe @xmath214 is a strict constraint for the form of @xmath133 .
it should be noted that the imaginary part of qcd coefficients @xmath173 only arises at the order of @xmath1 , and depends on the renormalization scale .
this dependence would bring some uncertainties in determining the @xmath3 asymmetries in @xmath0 decays .
maybe this scale dependence of the imaginary part could be cancelled by the results on higher order of @xmath1 .
comparing qcd factorization approach with the generalized factorization@xcite and pqcd method@xcite , some interesting comments are in order .
\(i ) at the zeroth order of @xmath1 , both of qcd factorization ( bbns approach ) and generalized factorization can reproduce the results of `` naive factorization '' ; at the higher order of @xmath1 , the renormalization scheme and scale dependence for the hadronic matrix elements can be recovered from the hard - scattering kernels @xmath216 in bbns approach and @xmath48 in generalized factorization . however , in generalized factorization , @xmath48 is from the one - loop calculations of quark - level matrix elements .
according to buras _
_ , quark - level matrix elements are accompanied with infrared divergences . to avoid these divergences
, one may assume that external quarks are off - shell .
unfortunately , it will introduce gauge dependence which is also unphysical .
but in the bbns approach , it is different because the external states are all physical and can be approximated as on - shell quarks in the leading order of @xmath13 . as a consequence
, the unphysical gauge dependence does not appear . in the pqcd method , li _
claim that their method is based on a six - quark system in which the external quarks are on - shell , so the gauge invariance of pqcd predictions is guaranteed .
the scale dependence in pqcd prediction is removed by evolving the wilson coefficients down to the proper hard scale .
so no explicit scale dependence is left .
\(ii ) in generalized factorization and bbns approach , the hadronic transition form factors are not calculable , and they are dominated by soft gluon exchange and determined only by experiments or some non - perturbative approaches such as sum rule , lattice qcd , etc . in particular , the above assumption can be also justified in the bbns approach by naive power counting @xcite .
however , in the pqcd method , this naive power counting rule may be invalid when the transverse momenta of the quark constituents and sudakov suppression are taken into account .
so , li _ et al .
_ thought that the hadronic transition form factors can be calculated in the pqcd method because @xmath8 quark is heavy enough and the soft gluon exchange is suppressed by sudakov form factors .
this is the essential difference between the bbns approach and the pqcd method .
\(iii ) the generalized factorization considers `` nonfactorizable '' contributions as intractable .
therefore , one may introduce one or more effective color numbers @xmath14 to phenomenologically represent `` nonfactorizable '' contributions @xcite .
furthermore , @xmath14 is assumed to be universal to maintain predictive power . however
, `` nonfactorizable '' contribution is really process - dependent . in the bbns approach and the pqcd method , such `` nonfactorizable '' contributions are indeed calculable in perturbative theory . in consequence
, @xmath14 need not be introduced .
\(iv ) as mentioned in the above sections , the strong phases predicted by generalized factorization are only from the bss mechanism which is represented by the penguin insertion .
however , the virtuality of the gluon or photon @xmath97 in the penguin insertion is ambiguous in generalized factorization , and usually it is approximated around @xmath98 .
this brings significant uncertainties for predicting the @xmath3 asymmetries for b decays .
a particular interesting result of the bbns approach is that strong phases are not only from the bss mechanism but also from the hard scattering , and there are no uncertainties in determining @xmath97 of penguin insertion .
however , compared with the real part of the decay amplitude , the imaginary part is @xmath217 or power @xmath13 suppressed and can not lead to large @xmath3 asymmetries , since they come solely from hard scattering processes which are only calculable in the heavy quark limit . in the pqcd method , there is no such @xmath1 suppression in the imaginary part of the decay amplitude .
thus @xmath3 asymmetries predicted by the pqcd method are usually greater than the prediction of bbns approach and generalized factorization .
so maybe these differences of prediction for @xmath3 asymmetries can be an experimental test for the bbns and pqcd approaches .
\(v ) hard spectator contributions [ fig .
1(g)and 1(h ) ] , which are leading power effects in qcd factorization , miss out in `` naive factorization '' and `` generalized factorization '' .
they are , however , @xmath218 suppressed compared with the leading factorized contributions ( the hadronic transition form factors ) .
but , in the pqcd method , they are of the same order as the form factors .
\(vi ) in the pqcd method , penguin contributions receive a dynamical enhancement called `` fat penguin''@xcite .
but in generalized and qcd factorization , they are missed .
this enhancement in the pqcd method arises from the strong scale dependence of the penguin wilson coefficients @xmath219 , etc .
\(vii ) final state interactions ( fsi ) do not appear in the three methods .
in qcd factorization , beneke _ et al . _
point out that the cancellation of the infrared divergences implies that the long distance fsi is power suppressed due to the quark - hadron duality . however , this point of view is controversial @xcite , but can be examined by the experimental measurements of @xmath220@xcite .
qcd factorization formula only holds in the heavy quark limit @xmath57 . in the real world @xmath44
is only about @xmath221 gev , the validity of the power suppression may be questionable .
in particular , for several cases , the power suppressed corrections can be numerically large@xcite , because the perturbative expansion is in order of @xmath1 which is not small at the realistic scale @xmath222 compared to @xmath13 .
\(i ) the hard `` nonfactorizable '' contributions computed by qcd factorization are generally small compared to the leading `` factorizable '' contribution .
but when the leading `` factorizable '' contributions are color suppressed , the `` nonfactorizable '' contribution may be larger than the leading results . at the same time , the potentially soft contribution , which is formally power suppressed , may be important .
for example , in @xmath223 , any perturbative and soft power suppressed contributions can have a significant effect on predicting the branching ratio and @xmath3 asymmetry .
furthermore , this problem also arises when the entire leading power contribution is suppressed by small wilson coefficients , for example , in @xmath224 ; or when the leading power contribution is suppressed by the small ckm elements .
\(ii ) an important power suppressed contribution is from the higher twist light - cone wave functions of the light mesons .
the chirally enhanced power correction from the two - particle twist-3 wave functions is the most important , and has been partly involved in this work except for the annihilation topologies .
other contribution from multiparticle non - valence fock state has been proved to be also power suppressed@xcite .
however , there is no systematic way to evaluate it .
the author of ref.@xcite proposed a way to evaluate the soft gluon exchange contribution from higher twist @xmath225 wave functions within the frame of the light - cone sum rule(lcsr ) .
but the accuracy of lcsr is limited due to the quark - hadron duality approximation . on the other hand , power correction from transverse
momenta needs a subtle treatment in the future . in ref.@xcite ,
the authors point out that the contribution from the transverse momenta might be considered when we evaluate the hadronic matrix elements to two - loop order . in this case , it is possible that sudakov suppression might be taken into account as well . in summary , up to now
, we do not have a systematic way to evaluate many kinds of power suppressed corrections for exclusive processes .
how to evaluate such corrections in a consistent way within the frame of qcd factorization is a potentially interesting work .
in this work , we give a detailed discussion for qcd factorization involving the complete chirally enhanced power corrections in the heavy quark limit for @xmath0 decays to two light pseudoscalar mesons , and present some elaborate calculations of radiative corrections at the order of @xmath1 .
we point out that the infrared finiteness of the vertex corrections in the chirally enhanced power corrections requires twist-3 light - cone distribution amplitudes ( lcdas ) of the light pseudoscalar symmetric .
however , even in the symmetric condition , there is also infrared divergence from the end point of the lcdas in the hard spectator scattering and annihilation topology .
so the transverse momenta and sudakov suppression should be taken into account .
we also point out that the decay amplitude of @xmath2 predicted by qcd factorization is really independent of the renormalization scale , at least at the order of @xmath1 . at last ,
we briefly compare the qcd factorization to the generalized factorization and pqcd method which are generally used in studying @xmath0 exclusive hadronic decays .
we thank professor hai - yang cheng and kwei - chou yang for pointing out errors in the coefficients of @xmath226 and @xmath227 in eqs.(50 ) and ( 52 ) .
and we also thank professor maozhi yang for helpful discussions about qcd factorization and the pqcd method .
this work is supported in part by national science foundation of china and state commission of science and technology of china .
two particle twist-2 and twist-3 light - cone distribution amplitudes of light pseudoscalar mesons are defined by the following nonlocal matrix elements@xcite : @xmath228 with @xmath229 being the decay constant of the light pseudoscalar , @xmath230 ( @xmath118 and @xmath119 are the masses of the constituent quarks in the pseudoscalar ) , and @xmath231 . here
@xmath95 is the twist-2 light - cone distribution amplitude ; @xmath139 and @xmath133 are two - particle twist-3 distribution amplitudes .
the above definitions can be combined into the below nonlocal matrix element : @xmath232 the renormalization group equation for the wilson coefficients @xmath24 is written as follows@xcite : @xmath234 here @xmath6 is the anomalous dimension matrix , which can be calculated by the perturbative theory and expanded in order of the coupling constants @xmath1 and
@xmath235 : @xmath236 the lo anomalous dimension matrix @xmath237 of the above equations has the explicit form @xmath238 where @xmath239 is the color number , @xmath240 is the active flavor number , and @xmath92 and @xmath83 denote the number of the active up- and down - type flavors respectively . in calculation of the perturbative diagrams shown in fig . 1
, one might encounter some feynman parameter integrals which involve nontrivial infrared divergence . to deal with the infrared divergence , as mentioned in preceding sections , the dimensional regularization ( dr ) and massive gluon ( mg ) scheme
are applied .
below , we give the explicit calculation of some useful feynman parameter integrals in the above two regularization schemes .
first , we deal with the integrals in dr scheme . in the dr scheme ( here we take @xmath89 and @xmath90 ) , the integrals involving the infrared divergence are written as follows : @xmath241 , \\
\int_0 ^ 1 dt_1 \int_{0}^{1-t_1 } dt_2 \frac{t_2}{(t_1(t_1+t_2 u))^{1-a}}&= & \frac{1}{u}\left[\frac{1}{a}-2+\ln u+\frac{\ln u}{1-u}\right ] , \\
\int_0 ^ 1 dt_1 \int_{0}^{1-t_1 } dt_2 \frac{(1-t_1)(1-t_2)}{(-t_1 t_2 u)^{1-a}}&=&-\frac{1}{u}\left [ \frac{1}{a^2}+\frac{\ln(-u)-2}{a}+\frac{27-\pi^2}{6}\right.\nonumber \\ & - & \left .
2\ln(-u)+\frac{1}{2}\ln^2(-u ) \right ] .
\end{aligned}\ ] ] @xmath244 , \\ & & \int_0 ^ 1 dt_1 \int_{0}^{1-t_1 } dt_2 \frac{t_2}{(t_1(t_1+t_2 \omega)+(1-t_1-t_2 ) \lambda)}\nonumber \\ & = & \frac{1}{u}\left[-\ln \lambda-1+\ln u+\frac{\ln u}{1-u}\right ] , \\ & & \int_0 ^ 1 dt_1 \int_{0}^{1-t_1 } dt_2 \frac{(1-t_1)(1-t_2)}{-t_1 t_2 u+(1-t_1-t_2)\lambda}\nonumber \\ & = & -\frac{1}{u}\left [ \frac{1}{2}\ln^2\lambda-(\ln(-u)-2)\ln \lambda - 2\ln(-u)+\frac{1}{2}\ln^2(-u)+\frac{5}{2}+\frac{\pi^2}{3 } \right ] . \end{aligned}\ ] ] a. abada et al .
( ape collaboration ) , phys .
b @xmath250 , 275 ( 1996 ) ; j.m . flynn and c.t .
sachrajda , _ heavy flavours ii _ , 402 - 452 , ( hep - lat/9710057 ) , j.m .
flynn et al.(ukqcd collaboration ) , nucl .
b @xmath251 , 327 ( 1996 ) ; l. del debbio et al .
( ukqcd collaboration ) , phys .
b @xmath252 , 392 ( 1998 ) .
a. khodjamirian , r. rckl , s. weinzierl and o. yakovlev , phys.lett.b @xmath254 , 275 ( 1997 ) ; a. khodjamirian , r. rckl and c.w .
winhart , phys.rev.d @xmath249 , 054013 ( 1998 ) ; a. khodjamirian , r. ruckl , _ heavy flavours ii _ , 345 - 401 , ( hep - ph/9801443 ) .
|
in this work , we give a detailed discussion for qcd factorization involved in the complete chirally enhanced power corrections in the heavy quark limit for @xmath0 decays to two light pseudoscalar mesons , and present some detailed calculations of radiative corrections at the order of @xmath1 .
we point out that the infrared finiteness of the vertex corrections in the chirally enhanced power corrections requires twist-3 light - cone distribution amplitudes ( lcdas ) of the light pseudoscalar symmetric .
however , even in the symmetric condition , there is also a logarithmic divergence from the endpoints of the twist-3 lcdas in the hard spectator scattering .
we point out that the decay amplitudes of @xmath2 predicted by qcd factorization are really free of the renormalization scale dependence , at least at the order of @xmath1 . at last
, we briefly compare the qcd factorization with the generalized factorization and pqcd method . _
/-2.5 mm _ * pacs numbers 13.25.hw 12.38.bx *
|
in this supplemental material we will first present photocurrent measurements . to confirm that a @xmath63 nm resonant
laser excites charges in the cavity region , we compare the photocurrent with when a @xmath53 nm off - resonant laser is used .
then we compare photoluminescence data with and without the presence of an aperture , to demonstrate that the oxide aperture plays an important role in trapping the charges , .
these observations support our claim that the mechanism behind the observed nonlinear effects in the resonant qd scans , is the excitation of charges that are trapped by the oxide aperture .
current as function of bias voltage was measured without ( dark ) and with the presence of a @xmath64 nm or a @xmath63 nm laser , see fig .
[ supplfig1 ] ( a ) .
the @xmath63 nm laser frequency was locked at the cavity resonance , such that the absorption is increased due to the high finesse ( @xmath65 ) of the cavity , as is the case during the resonant scans in the main text .
when no additional pump laser is applied , a current flows through the device in the same direction as that the forward bias voltage is applied , giving rise to an iv - curve that is typical for a pin diode device . in the presence of an additional pump laser charges
are now optically excited and flow in the opposite direction due to the internal field being present , giving rise to an additional negative photocurrent . in fig .
[ supplfig1 ] ( b ) we compare the absolute value of the photocurrent produced by the @xmath66 nm and @xmath53 nm laser , normalized by the pump laser intensity . for both lasers a strong bias voltage dependency is visible and the collected photocurrent is smaller at a larger bias voltage , which is closer to the flatband regime .
this indicates that either the absorption , or the collection efficiency of the excited charges , is voltage dependent .
the 940 nm laser gives rise to a photocurrent of about 0.5 ma / w at a bias voltage of 700 mv where charge neutral qds are typically operated , which at an intensity of @xmath67 nw corresponds to about @xmath68 charges being excited and collected per second .
next , we compare the photoluminescence collected from the oxide - aperture microcavity and from a region with an unbalanced cavity . in the unbalanced cavity region
the top dbr has been nearly completely etched away , in order to enhance the collection efficiency , and no oxide aperture has been applied .
figure [ supplfig2 ] ( a , c ) present colormaps of the pl spectra as a function of applied bias voltage and collected wavelength , using a @xmath53 nm pump laser of various intensities . figure [ supplfig2 ] ( b ) shows the pl intensity as function of voltage at the cavity resonance .
single qd lines ( marked by the arrows ) are hardly distinguishable from the background cavity emission , which originates from broadband wetting layer emission . for increasing pump laser intensity ,
it is clearly visible that the qd lines appear at a lower bias voltage . for the qds in the region without an oxide aperture ,
the shift in voltage is however much smaller .
the qds in fig .
[ supplfig2 ] ( c ) are now better visible , even though at high pump laser intensities for increasing voltage again broadband emission washes out clear qd signatures .
figure [ supplfig2 ] ( d ) presents the mean pl counts for various pump intensities .
a shoulder is visible , which corresponds with the voltage range where the qds are on average charge neutral
. the shoulder does shift to a lower voltage range , but the shift ( in total @xmath69 mv for 1100 @xmath55w ) is much smaller compared to the total @xmath70 mv ( for 1500 @xmath55w ) voltage shift of the qds in the microcavity .
even though it is likely that roughly the same amount of charges are excited by the @xmath64 nm laser in the microcavity region compared to elsewhere in the sample , these results indicate that the oxide aperture provides a charge memory , causing a much larger effective charge buildup resulting in a larger voltage shift of the qds .
31ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1038/nphys708 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.093604 [ * * , ( ) ] http://view.ncbi.nlm.nih.gov/pubmed/23003653 [ * * ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.109.166806 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2012.181 [ * * , ( ) ] link:\doibase 10.1126/science.1154643 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2014.114 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.108.183601 [ * * , ( ) ] link:\doibase 10.1038/nature07127 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.160503 [ * * , ( ) ] http://journals.aps.org/rmp/abstract/10.1103/revmodphys.85.79 [ * * , ( ) ] link:\doibase 10.1038/nmat3652 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.108.107401 [ * * , ( ) ] link:\doibase 10.1038/nphys2688 [ * * , ( ) ] link:\doibase 10.1063/1.1999843 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2007.227 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.097403 [ * * , ( ) ] link:\doibase 10.1063/1.3276550 [ * * , ( ) ] link:\doibase 10.1063/1.4872003 [ * * , ( ) ] http://journals.aps.org/pra/abstract/10.1103/physreva.75.053823 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.96.153601 [ * * , ( ) ] http://www.amazon.com/exec/obidos/redirect?tag=citeulike07-20&path=asin/0471143235[__ ] , ed .
( , ) link:\doibase 10.1063/1.118103 [ * * , ( ) ] link:\doibase 10.1063/1.118897 [ * * , ( ) ] link:\doibase 10.1063/1.113489 [ * * , ( ) ] http://journals.aps.org/prx/abstract/10.1103/physrevx.4.021004 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.245306 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.98.026806 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.99.056804 [ * * , ( ) ] http://journals.aps.org/prb/abstract/10.1103/physrevb.74.081306 [ * * , ( ) ] link:\doibase 10.1038/nphys1363 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.108.197403 [ * * , ( ) ] see supplemental material at [ url will be inserted by publisher ] for details on photocurrent measurements with lasers of different wavelengths , and a comparison between photoluminescence measured at the microcavity and measured in a region without oxide aperture .
|
in an oxide apertured quantum dot ( qd ) micropillar cavity - qed system , we found strong qd hysteresis effects and lineshape modifications even at very low intensities corresponding to @xmath0 intracavity photons .
we attribute this to the excitation of charges by the intracavity field ; charges that get trapped at the oxide aperture , where they screen the internal electric field and blueshift the qd transition .
this in turn strongly modulates light absorption by cavity qed effects , eventually leading to the observed hysteresis and lineshape modifications .
the cavity also enables us to observe the qd dynamics in real time , and all experimental data agrees well with a power - law charging model .
this effect can serve as a novel tuning mechanism for quantum dots .
cavity quantum electrodynamics with quantum dots ( qds ) coupled to microcavities enables various applications such as single - photon switches @xcite , generation of non - classical states of light @xcite and hybrid quantum information schemes @xcite . however , qds deviate from an ideal atom - like systems as they strongly interact with their environment , for example through nuclear spins @xcite and via charge traps @xcite .
these interactions need to be understood and controlled in order to improve the qd coherence properties . for this purpose cavities
are very useful to probe the qd environment , through increased light - matter interaction . and reflectivity @xmath1 . the qd cooperativity @xmath2 and dephasing rate @xmath3 obtained from the fits are named in the figures .
the black arrows denote a second qd in the same cavity .
scans were taken on @xmath4s timescale .
( d ) presents simulations that predict the scans in ( c ) . ] in this letter we investigate such a qd - cavity system . for sufficiently low optical field intensity
this system can be described by the qed of an effective 2-level system in a single - mode cavity .
for increasing intensities we report on bistable and strong nonlinear behavior .
the sample under study consists of inas self - assembled qds inside a pin diode structure embedded in a micropillar .
this system combines qd charge and stark shift control by applying a bias voltage with high - quality polarization - degenerate cavity modes @xcite .
the mode confinement in the transversal direction is achieved by an oxide aperture formed through a wet oxidation step .
the observed bistability and nonlinear behavior in the cavity qed system can be explained by attributing a second role to the oxide aperture , namely that of a charge memory .
charges in this memory , created by resonant absorption , will cause a modification to the applied bias voltage which in turn shifts the qd frequency , and modify the amount of absorption . in figure [ fig1 ] ( a ) the sample structure , with charges trapped at the oxide aperture ,
is schematically shown .
we consider one of the fine - split transitions of a charge neutral qd coupled to a polarization degenerate cavity mode in the intermediate coupling regime .
figure [ fig1 ] ( b ) shows reflection spectra , recorded at a sufficiently low incident intensity @xmath5 pw such that no nonlinear effects occur .
upward and downward frequency scans overlap perfectly and can be fitted by theory for a dipole inside an optical cavity , which we will discuss later in detail . however , when a higher intensity of 1 nw is used , several strong deviations occur , see fig .
[ fig1 ] ( c ) .
first of all , a hysteresis feature appears when the qd is tuned at or below the cavity resonance .
second , while at the high frequency side of the cavity resonance the hysteresis is much less , a line shape modification is still visible .
finally , in order to obtain the same qd detuning compared to the low intensity scans , a lower bias voltage has to be applied .
after a thorough characterization of this effect , we present a model that explains all features as is shown in fig .
[ fig1 ] ( d ) . a first hint on the underlying dynamics is provided by investigating the power dependence in single - laser scans . in fig .
[ fig2 ] ( a ) we first keep the qd bias voltage constant and show scans for increasing laser intensity .
a qd blueshift occurs , as is displayed in fig .
[ fig2 ] ( c ) .
in a second set of measurements in fig . [ fig2 ]
( b ) , we keep the qd - cavity detuning constant by changing the bias voltage as function of the laser intensity .
a lower bias voltage has to be applied for increasing laser intensity , shown in fig .
[ fig2 ] ( d ) .
furthermore , the hysteresis width increases with intensity , see fig .
[ fig2 ] ( e ) , up to nearly 6 ghz , but then saturates and even decreases slightly when the intensity is above the qd saturation intensity ( @xmath6 nw ) .
all three observations in figs . [ fig2 ]
( c - e ) obey the same empirical power - law @xmath7 , with @xmath8 .
already at an incident intensity @xmath9 pw , corresponding to a maximum mean photon per cavity lifetime of @xmath10 , a qd blueshift and line shape modification is clearly visible . here
the mean intracavity photon number is found from @xmath11 , where @xmath12 ns@xmath13 is the mirror loss rate , @xmath14 is the light angular frequency , and the maximum output intensity @xmath15 pw ( see below ) .
( c ) that have been added as a reference .
vertical offsets are added to the curves . ]
we now switch to a two - laser experiment ( see fig . [ fig3 ] ) in order to further investigate the phenomenon that the qd blueshifts with increasing laser intensity .
the qd transition is probed with a low intensity ( 1 pw ) probe laser such that no line shape modification occurs .
we then add a second high intensity ( 1 nw ) pump laser with orthogonal polarization , such that it can be filtered out with a crossed polarizer in the detection channel .
the pump laser is scanned in steps across the cavity resonance ; for each step the qd - cavity spectrum is measured with the weak probe laser .
figure [ fig3 ] ( a - c ) presents these measurements for various bias voltages such that the average qd - cavity detuning is varied .
for every bias voltage the qd - cavity and pump laser - cavity detunings are determined , and the relative qd frequency as function of the pump laser - cavity detuning is shown in fig .
[ fig3 ] ( d ) . the maximum qd blueshift ( of about 8 ghz ) occurs when the pump laser is resonant with the cavity mode , corresponding to the vertical dashed line .
an increased blueshift also occurs when the pump laser is close to the qd frequency , indicated by the arrows in fig .
[ fig3 ] ( d ) , at which point the intracavity field also increases due to cavity qed effects .
the gray dashed curves are lorentzian lines convoluted with the @xmath16 power - law from fig .
[ fig2 ] ( c ) and correspond to the data nicely . in conclusion , the data clearly shows that the qd resonance blueshifts when the intracavity field increases , and this effect is independent of the qd - laser or the qd - cavity detuning .
we now introduce a model that explains the dynamic line shape modifications .
we start from the transmission amplitude of a cavity with a coupled dipole @xcite : @xmath17 where @xmath18 is the relative detuning between the laser ( @xmath14 ) and cavity ( @xmath19 ) angular frequencies , @xmath2 is the device cooperativity , @xmath20 is the relative detuning between the laser and qd transition ( @xmath21 ) , @xmath22 is the output coupling efficiency , @xmath23 is the total intensity damping rate of the cavity and @xmath3 is the qd dephasing rate . we obtain close to perfect mode - matching , and therefore the total transmittivity through the cavity is given by @xmath24 , and the total reflectivity is given by @xmath25 . in fig .
[ fig1 ] ( a ) we show that the model fits the low intensity measurements very well .
the qd cooperativity and dephasing parameters @xmath2 and @xmath3 are noted in the subfigure windows , and a cavity damping rate @xmath26 ns@xmath13 and an output coupling efficiency @xmath27 is found , corresponding to a cavity @xmath28-factor of @xmath29 . as a next step ,
we introduce a qd frequency @xmath30 that dynamically changes with the intracavity intensity , which is proportional to @xmath31 : @xmath32 where @xmath33 equals the qd frequency in the limit of vanishing intracavity intensity . based on the empirical values determined in fig .
[ fig2 ] ( c ) , we use @xmath34 ghz , with @xmath35 in pw , and @xmath36 . due to cavity qed effects
, the intracavity field depends strongly on the qd frequency such that the cavity can change from being largely transparent to being largely reflective through only small changes in the qd frequency .
this interplay leads to the observed nonlinear behaviour . finally , we take into account that we are operating the qd cavity system close to the saturation intensity , which slightly suppresses the qd features .
we take this into account by calculating the reflectivity @xmath37 by taking the weighted sum of the reflectivities of a coupled ( @xmath38 ) and uncoupled ( @xmath39 ) cavity : @xmath40 , with @xmath41 .
this is a strong simplification of a more rigorous approach based on the quantum master equation and calculation of the intracavity photon number @xcite , but is sufficient for our purpose . for qd cooperativity @xmath2 and dephasing rate @xmath3
we use the values obtained in the fits in fig .
[ fig1 ] ( a ) .
figure [ fig1 ] ( c ) shows that the predicted scans match the actual measurements very well . as the underlying physical mechanism
we hypothesize that charges are excited by the resonant laser and get trapped at the oxide aperture .
reasonably high doping concentrations were used , up to @xmath42 @xmath43 for the carbon p - doped layers and up to @xmath44 @xmath43 for the silicon n - doped layers , and it is well - known that absorption takes place in these doped layers @xcite , resulting in a non - neglible photocurrent ( see the supplemental material section i @xcite ) . the excited charges partly screen the internal electric field responsible for the quantum confined stark effect in the qd , leading to the observed blueshift of the qd transitions .
furthermore , the fact that even for a very low cavity mean photon number of @xmath45 ( @xmath46 pw ) nonlinearities take place , indicates that the oxide aperture must form an efficient charge memory compared to the qd - cavity decay rate ( 77 ns@xmath13 ) .
it is well known that the interface between gaas and aluminum oxide ( alo@xmath47 ) , produced by wet oxidation of alas , provides a very high density of charge traps , in the form of amorphous oxide and micro crystallites @xcite , and in the form of elemental interfacial as @xcite , leading to spatially non - uniform fermi level pinning @xcite .
the time - resolved charge decay measurements presented below confirm that the charges are relatively long - lived on a @xmath48 ms timescale .
also , a comparison with qds observed outside of the oxide aperture region show that laser induced blueshift is neglible in these regions ( see supplemental material section ii @xcite ) .
furthermore , the charging hypothesis agrees with the observed @xmath49 sublinear power - law , which would be the consequence of coulomb repulsion between trapped charges , possibly in combination with increased filling of trap states that have an increasing decay rate .
our results agree with observations in ref .
@xcite , where single charge fluctuations are probed at a gaas / alas interface located about 50 nm away from the qds , while a gradual variation was observed for charge traps more than 150 nm away .
we did not observe any single
charge influences as the distance between the qds and the oxide aperture is about 200 nm in our sample .
in contrast to our work , the gaas / alas interface studied in ref .
@xcite is located in the intrinsic region , while the aperture in our sample is located in the p - doped region , but a similar qd blueshift for increasing laser intensity and constant bias voltage was observed .
nm ) laser .
( a ) shows the bias voltage where the qd transition is resonant with the cavity mode , as function of the intensity @xmath50 of the off - resonant laser .
the black dashed line indicates the qd bias voltage when no off - resonant laser is applied .
( b ) example of resonant reflection scans using a weak ( 1 pw ) probe laser , recorded in the presence of an off - resonant laser .
text in the figures denotes the applied bias voltage @xmath51 , intensity @xmath50 , and qd cooperativity @xmath2 and dephasing rate @xmath3 of the predicted red line .
( c , d ) colormaps of the reflected intensity of the probe laser fixed to the center of the cavity resonance ( light / dark : high / low signal , corresponding to an on(off ) resonant qd ) , as function of the time and applied bias voltage , for various off - resonant pump intensities . the off - resonant laser is turned on ( c ) and off ( d ) at @xmath52 ms .
the red line in ( a ) and the black - white dashed in ( c , d ) are reproduced using the same parameters ( see main text for explanation ) .
note that the time axes of ( c , d ) are different .
] finally , we performed a set of measurements to gain insight in the temporal dynamics of the charge build - up and decay .
for this purpose we use a laser ( @xmath53 nm ) that is off - resonant with the cavity but is resonant with the wetting layer . at a larger laser intensity
, more charges are excited compared to the resonant laser and larger qd shifts can be obtained .
we now use the coupled qd - cavity system as a very sensitive probe of the internal electric field , a principle that was also used to monitor a single charge trap in real - time @xcite .
figure [ fig4 ] ( a ) shows , for various off - resonant pump laser intensities , the bias voltage that has to be applied to tune the qd to the cavity resonance , determined by using a weak ( @xmath41 pw ) resonant probe laser .
again , as in fig .
[ fig2 ] ( d ) , a clear sub - linear behavior @xmath54 is visible . at an even higher intensity of 100 @xmath55w , a bias voltage of 350 mv had to be applied to tune the qd in to resonance , clearly indicating that the charge buildup and qd blueshift is not easily saturated .
strikingly , even for an excitation power of 200 nw , the qd has the same cooperativity @xmath2 and dephasing rate @xmath3 as when no off - resonant laser is present , see fig .
[ fig4 ] ( b ) .
this indicates that , even though many charges are excited , they are located relatively far away and give rise to a more - or - less constant effective electric field , thereby preserving the qd coherence . in order to directly monitor the time dynamics of the charge buildup and decay , we fix a weak probe laser that will not excite any additional charges at the cavity resonance , and monitor the reflectivity as function of bias voltage and time .
figure [ fig4 ] ( c ) shows , for various off - resonant pump laser intensities , reflectivity colormaps of the probe laser when the off - resonant pump laser is turned on at @xmath52 ms .
for @xmath56 ms , the qd is resonant with the cavity resonance at @xmath57 mv ( white dashed lines ) , corresponding with a high probe laser reflectivity at this voltage and a low reflectivity at different voltages .
the reflectivity at @xmath57 mv abruptly decreases as the pump laser is turned on , but is then restored at lower bias voltages , demonstating direct probing of the charge build - up . for increasing pump laser intensities ,
the reflectivity is restored at a lower voltage , but also the time for the charge build - up to reach an equilibrium decreases . the decay of the charges is monitored in fig .
[ fig4 ] ( d ) , where the pump laser is turned off at @xmath52 ms . for @xmath56 ms the qd reflectivity signal is now highest for a low bias voltage when the charge reservoir is saturated .
after the pump is turned off and the charges disappear , the reflectivity now gets restored at an increasing bias voltage .
the charge decay rate is initially fastest , when many charges are still present and the qd is resonant at a lower bias voltage , but then strongly decreases and finally occurs on a @xmath410 ms timescale , much slower than the charge build - up rate .
we now introduce a simple power - law model to describe the charge build - up and decay dynamics .
we assume that the qd voltage shift @xmath58 is proportional to the number of trapped charges @xmath28 , and increases as a function of the pump laser intensity @xmath59 , such that @xmath60 . as a result
, the charge decay and buildup is described using : @xmath61 .
the red line in fig .
[ fig4 ] ( b ) and the white - black dashed lines fig .
[ fig4 ] ( c - d ) are reproduced with @xmath62 and the same power - law scaling factor @xmath36 as found earlier , and describe the data nicely .
the effect we found enables a novel method to tune qd transitions without the need for electrical contacts , or enable independent tuning of qds sharing the same voltage contacts .
furthermore it could serve as a low power all - optical switch mediated by a charge memory .
the cavity - enhanced feedback mechanism with the charge environment could in principle also occur in other solid - state microcavity structures , where doped layers are present in which charges can be excited and where material defects or interfaces could act as a charge trap .
this interaction therefore also has to be taken into account when studying dynamical nuclear spin polarization ( dnp ) effects in a cavity , which is of general interest to prolong the qd coherence time and the potential to form a quantum memory .
dnp gives rise to qd line shape modifications , hysteresis and bistabilitity behavior @xcite ; phenomena that could also give rise to cavity - enhanced feedback mechanisms . in conclusion ,
we have studied a neutral qd transition coupled to a microcavity and observed strong cavity - enhanced feedback with the charge environment .
hysteresis and modifications of the qd line shape are demonstrated at intensities of way less than 1 photon per cavity lifetime , which we explain and model by a qd frequency blueshift attributed to charges trapped at the oxide aperture as a function of the intracavity field .
in general , these results demonstrate the potential of studying and controlling the qd environment using a cavity qed system .
this work was supported by nsf under grant no . 0960331 and 0901886 and fom - nwo grant no .
08qip6 - 2 .
|
variable resistors are commonly used circuit elements in many electronic applications .
however , their large size and slow response time have heretofore limited their use primarily to the human - circuit interface . in this article , we describe how the exploitation of quantum and stochastic effects at the nanoscale @xcite allows one to combine what would ordinarily be distinct macroscale circuit elements into a single nanoscale device with multiple functionalities , and to achieve response times on the order of picoseconds .
the device architecture we propose is illustrated in fig .
[ fig : diagram ] .
the physics behind its operation is the following : a metal nanowire is the active circuit element , and is embedded in a dielectric sheath , surrounded by an outer conductor of comparable dimensions , referred to as the gate .
a positive / negative voltage applied to the gate enhances / depletes the density of carriers in the nanowire .
this results in a shift of the electronic fermi energy @xmath2 , which alters the electron - shell structure of the nanowire @xcite .
this in turn changes the energy barriers that determine the rates of stochastic transitions @xcite between different conducting states of the nanowire .
the idea is that a sufficient voltage applied at the gate will make the energy barrier small enough to be comparable to the thermal energy @xmath3 , thus making a stochastic transition all but certain .
such a device would require as active element a nanowire where electron - shell effects dominate over atomic - shell effects @xcite .
this has been shown @xcite to be the case for wires of conductance between a few and about a hundred conductance quanta @xmath1 .
such nanowires with lengths below or around a few nanometers have been fabricated using various techniques , including scanning tunneling microscopy ( stm ) @xcite , mechanically - controllable break junctions ( mcbj ) @xcite , thin - film transmission electron microscopy ( tem ) @xcite , electromigration @xcite , and electrochemical fabrication @xcite .
nanowires with diameters less than a nanometer have been directly observed @xcite using tem to remain stable under low beam intensities below @xmath4 for the duration of observation .
stochastic switching between different conductance values has been observed in contacts made using mcbj @xcite , while controllable switching has been achieved recently using electromigration to grow or shrink a nanobridge between two wires @xcite .
a structural thinning process of the nanowire similar to the one described by the theory of @xcite @xcite has been observed to take place for gold nanowires in tem experiments .
the nanowire was observed @xcite to thin step by step via a process where a structural step ( corresponding to a change in radius of the order one atomic diameter ) forms at one end of the wire and subsequently propagates along the wire .
a nanoelectromechanical switch similar to the device we are proposing has recently been built by @xcite on a somewhat larger scale .
however , the larger scale of their device results in a different physical process responsible for the switching . @xcite argue that the switching in their device is the result of a bending of the wire due to an electrostatic attraction between the wire and the gate
. such bending would be negligible in the device we envision , which would be much shorter ( few nm as opposed to close to a @xmath5 m in ) , and the switching would be instead driven by stochastic events as described above .
the different scales , configurations , and physics underlying the operation of the two leads to different device characteristics and functionalities ; these will be reviewed at the end of this article .
metal nanowires have attracted considerable interest in the past decade due to their remarkable transport and structural properties @xcite .
long gold and silver nanowires were observed to form spontaneously under electron irradiation @xcite , and appear to be surprisingly stable .
even the thinnest gold wires , essentially chains of atoms , have lifetimes of the order of seconds at room temperature @xcite .
metal nanowires exhibit striking correlations between their stability and electrical conductance @xcite . that these filamentary structures are stable at all is rather counterintuitive @xcite , but can be explained by electron - shell effects @xcite . because most of their atoms are at the surface , with low coordination numbers , metal nanowires behave essentially like fluids @xcite .
classically , the rayleigh instability would break up any wire whose length exceeds its circumference @xcite .
nevertheless , nanowires clearly violating the rayleigh criterion have been observed repeatedly.@xcite the instability is suppressed through quantum effects , with stabilization occurring through the nanowire s electronic shell structure .
a quantum linear stability analysis @xcite showed the existence of `` islands of stability '' for discrete intervals of the radius @xmath6 .
these correspond to conductance `` magic numbers '' that agree with those observed in experiments . for low enough temperatures ,
there remain finite regions of @xmath6 stable against long - wavelength perturbations .
therefore , stable wires exist only in the vicinity of certain `` magic radii '' and consequently at quantized conductance values @xmath7 that are integer multiples of the conductance quantum @xmath1 .
however , the linear stability analysis ignores large thermal fluctuations that can lead to breakup of the wire .
nanowire lifetimes are inferred from conductance histograms , compiled by cycling a mcbj thousands of times @xcite .
these studies indicate that conductance peaks disappear above fairly well - defined temperatures , suggesting a lifetime at that temperature that is comparable to a typical wire - elongation time in the experiment . @xcite studied the lifetimes of these nanowires using techniques developed in . by modeling thermal fluctuations through stochastic ginzburg - landau classical field theories
, they constructed a self - consistent approach to the fluctuation - induced `` necking '' of nanowires that is in good agreement with experiment @xcite .
their theory indicates that passivated noble metal nanowires are sufficiently stable at room temperature to serve as interconnects between nanoscale circuit elements . of particular interest for the applications considered in this proposal
is the nature of the barriers separating wires of different magic radii ( and hence quantized conductances ) .
these barriers can be surmounted in several ways : among them are raising the temperature , applying strain , shortening the wire , or changing the fermi energy .
the first three are discussed in , but the last is new to this paper . for the purposes of a new nanoscale device , the first two may be unsatisfactory for various reasons having to do either with nonoptimal operating conditions ( temperature ) , or probable inability to implement these controls on the nanoscale ( strain ) .
the conductance of the wires with `` magic radii '' is indicated on the top axis . from , width=321 ] .
[ fig : shell_potential ] the magic radii are the minima of the electron - shell potential ( see fig . [
fig : shell_potential ] , and e.g. ) , which depends on the dimensionless parameter @xmath8 , with @xmath9 the fermi wavevector and @xmath6 the wire radius .
a shift in @xmath9 is thus analogous to applying strain , and can be used to induce rapid ( i.e. , on the scale of the debye frequency ) transitions between neighboring magic radii . these have conductances differing by @xmath10 , where @xmath11 is an integer ( see fig . [
fig : shell_potential ] ) .
( as a rule of thumb , the jumps scale as @xmath12 for a wire with initial radius @xmath6 . )
the switching time between two adjacent magic radii was shown @xcite to be given by the kramers formula @xmath13 where @xmath14 is the energy barrier , @xmath15 is boltzmann s constant , and @xmath16 is the temperature .
the rate prefactor @xmath17 , of order the debye frequency , was calculated explicitly in .
the dependence of @xmath14 on the parameter @xmath8 is illustrated in fig .
[ fig : strain ] . , or equivalently applied stress @xmath18 .
results correspond to a gold wire with a conductance @xmath19 , following the calculations of .
, width=321 ] the possibility of shifting @xmath2 electrostatically , as described above , depends in an essential way on the crucial feature that the nanowire has a radius of order nanometers , and thus has a very low density of states at @xmath2 . as a function of the applied gate voltage @xmath20 ,
the shift in @xmath2 is given by @xcite @xmath21 where @xmath22 is the mutual capacitance between gate and nanowire , and @xmath23 is the density of states of the nanowire at @xmath2 . as discussed in , the denominator in eq .
[ eq : de_f ] can be well approximated in terms of material and geometrical parameters , yielding a convenient rule - of - thumb estimate @xmath24 where @xmath25 is the fermi gas parameter for the nanowire material ( essentially the mean inter - electron separation in the bulk metal ) , @xmath26 is the mean dielectric constant of the dielectric sheath , and @xmath27 is a dimensionless parameter of order unity , which depends logarithmically on the device dimensions . in order to achieve the maximum switching speed , it is necessary to achieve a shift @xmath28 in the shell - potential parameter . from eq .
[ eq : de_f_est ] , this implies a preferred operating gate voltage @xmath29 for typical metals , @xmath303 , while @xmath31 in the domain of validity of the nanoscale free electron model @xcite .
it is therefore desirable to use a dielectric with @xmath32 to minimize the necessary gate voltages .
because the mechanical switching time of the nanoscale variable resistor can be as short as picoseconds , it may also be thought of as an _ electromechanical transistor_. it is thus useful to compute its _ transconductance _ , a figure of merit used to characterize transistors .
the transconductance @xmath33 can be estimated as @xmath34 where @xmath35 is the device bias voltage .
using eq .
[ eq : dvg_est ] and @xmath12 , one finds @xmath36 for large dielectric constants @xmath37 , and bias voltages @xmath38v , one can thus achieve @xmath39 , an exceptional figure of merit for a nanoscale device @xcite , thereby enabling its advantageous use as an effective transistor .
in addition to the structural switching time of order picoseconds , the electrical @xmath40 rise time @xmath41 can be estimated to be of order 1 femtosecond for typical device dimensions / materials , and so is not a limiting factor in device performance .
( top ) and prefactor @xmath42 ( bottom ) as a function of wire length , as calculated in a second - order phase transition takes place at the critical length @xmath43 . for some wires ,
the transition is first - order ( inset).,width=321 ] the device discussed above is one where barriers are controlled by shifting the fermi energy of the nanowire through electrostatic means .
another possibility is to change the wire length .
it was predicted that a transition in activation behavior occurs as a function of wire length : below a critical length @xmath43 , the barrier decreases rapidly with length , while above it is roughly constant .
the transition can be continuous ( second - order ) or discontinuous ( first - order ) ( see fig .
[ fig : transition ] ) .
this effect may have already been observed : a recent study @xcite reported a transition from linear to nonlinear @xmath0 behavior in gold nanowires , as distance between electrodes shortened due to an applied bias ( presumably because of thermal expansion of the electrodes ) . in a comment @xcite , we were able to explain this change in @xmath0 behavior as a consequence of the transition in radius stability as a function of wire length ( cf .
[ fig : transition ] ) .
this suggests another device possibility , namely one where the wire length can be shortened below its critical value directly by increasing the applied voltage . this would convert a wire with linear @xmath0 characteristics to one with nonlinear ones ( see fig .
[ fig : iv ] ) . at the present time , it remains unclear how easily controllable such a transition might be and whether it would be reversible .
construction of a nanoscale variable resistor / electromechanical transistor in the laboratory will require combining the three components of the device
metal nanowire , dielectric , and gate in a single nanostructure .
to achieve optimal device characteristics , the space between the active segment of the nanowire and the gate(s ) should be filled with a dielectric with @xmath32 .
if a solid dielectric ( only ) is used , a small gap around the active segment of the nanowire must be provided ( see fig . [
fig : pit](b ) ) to permit the nanowire surface to fluctuate freely . in that case , the mean dielectric constant of the region between the nanowire and the gate(s ) ( including the gap ) should exceed ten .
many intrinsic semiconductors could serve as suitable solid dielectrics with @xmath32 ( e.g. , si , ge , insb , inas , inp , gasb , or gaas ) .
the material should be chosen so that the semiconducting energy gap exceeds the maximum desired voltage difference between the gate and nanowire . a liquid dielectric or combination of solid and liquid dielectrics could also be utilized .
this would allow for optimal filling of the dielectric region , while still permitting free motion of the nanowire surface .
liquid dielectrics have been used in conjunction with some of the previous techniques @xcite , in the context of single molecule measurements , as well as for stm measurements of metal contact transport . in the latter context , they have been shown to have little influence on the stability and transport properties of the nanocontact @xcite .
below we propose two architectures to realize the device described above . in a first possible architecture for the proposed device ( see fig . [
fig : droplet ] ) , a layer of solid dielectric is deposited on a substrate prepatterned ( using standard vapor deposition techniques ) with a metallic gate to address the nanowire device .
a metal nanowire several tens of nanometers in diameter with a `` notch '' or constriction at the desired location is then deposited on the surface of the dielectric , in alignment with the submerged gate electrode .
the nanowire at the notch can then be thinned down to the specified operating diameter by electromigration , scanning - electron microscopy ( sem ) , or chemical etching .
a nanodroplet of liquid dielectric is then deposited on the surface of the wafer , immersing the nanowire device ( see fig . [
fig : droplet ] ) . the nanowire device , together with the droplet of liquid dielectric ,
is then hermetically sealed , e.g. , with an epoxy seal . in another realization of the proposed device ( see fig . [
fig : pit ] ) , a layer of solid dielectric is deposited on a substrate prepatterned with a metallic gate to address the nanowire device , as in the first realization .
a metal nanowire of uniform diameter several tens of nanometers is then deposited on the surface of the dielectric , in alignment with the submerged gate electrode .
this fabrication step can be carried out within standard semiconductor patterning techniques , such as for example e - beam direct write or alternatively in the long term masked ion beam lithography .
a further layer of solid dielectric is then deposited , fully encasing the nanowire .
a nanoscale pit or cavity is then etched in the dielectric layer , exposing the active segment of the nanowire .
the exposed segment of the nanowire is then thinned down to the specified diameter ( of order one nanometer ) via e.g. , focused sem , chemical etching , or electromigration , or a combination of these techniques .
a hermetic seal can be applied to increase the durability of the nanowire device .
for example , an epoxy bubble seal may be used to enclose an inert atmosphere ( e.g. , nitrogen or argon ) about the exposed segment of the nanowire .
alternatively , a passivation layer over the nanowire device is used to scavenge any small amounts of oxidant from the sealed environment .
the pit containing the nanowire ( see fig . [
fig : pit](b ) ) can also be filled with a liquid or gel dielectric before the seal is applied , to enhance device performance . for some applications , both a top gate ( not shown ) and a bottom gate
are included , above and below the nanowire device , respectively .
multiple gates may be desirable e.g. to address individual devices in a large array . for example
, if the gate voltage is chosen appropriately , the device will switch conducting states rapidly only if the voltage is applied to both gates . because the three terminals of the nanoscale variable resistor consist of metal which can be patterned by standard semiconductor fabrication techniques , such devices can be readily integrated with conventional circuitry on a chip . because the throughput impedance of such a device is on the scale of several hundred to several thousand ohms
, appropriate amplification may be required to interface with standard cmos circuitry .
we have proposed a nanoscale device that exploits the new physics that emerges at the boundary between classical and quantum physics . a monovalent metallic nanowire is stable at certain `` magic radii '' @xcite , corresponding to a subset of integer multiples of the quantum conductance @xmath1 .
the nanometer lengthscale of these systems leads to low energy barriers between different conductances that are surmountable through classical thermal fluctuations on short timescales .
the barriers themselves are functions of both the wire s material parameters for example , electron density of states and wire geometry , and as such are controllable through external manipulation .
this leads to a novel situation where a basic wire configuration can perform multiple functions that on the macroscopic scale require different specially designed electronic components .
the configuration of the device a metal nanowire embedded in a dielectric sheath surrounded by an outer conductor , which serves as the gate is simple . changing the gate voltage alters the density of carriers , which in turn allows sensitive control of the energy barriers that determine the rates of stochastic transitions between different conducting states of the nanowire @xcite .
the device can alternatively serve as a simple resistive element , as a variable resistor , or as an electromechanical transistor , given a switching time that can be of the order of picoseconds .
a further possibility , though one that may be more difficult to control , is using external voltage to change between linear and nonlinear @xmath0 response , through a transition in activation behaviors as noted in .
the device has some similarities with one built by @xcite , but there are important differences .
the latter , which is realized on the @xmath45 m rather than the nm lengthscale , depends on a mechanical process ( wire bending ) for its operation , leading to a slower response which effectively reduces the array of functionalities of the device . on the other hand
, the device of can serve as an on - off switch , whereas the device discussed here can change only between different nonzero conductances .
the two devices are therefore best viewed as complementary , with each specialized to different but useful functions .
this work was supported by nsf grant nos .
0312028 ( cas ) , and phy-0651077 and phy-0965015 ( dls ) .
part of this work was done when cas and dls were at the aspen center for physics , and jb was at the university of arizona .
the authors acknowledge the hospitality of the santa fe institute where the last stages of this work were completed .
37ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( )
|
a nanoscale device consisting of a metal nanowire , a dielectric , and a gate is proposed .
a combination of quantum and thermal stochastic effects enable the device to have multiple functionalities , serving alternately as a transistor , a variable resistor , or a simple resistive element with @xmath0 characteristics that can switch between ohmic and non - ohmic . by manipulating the gate voltage ,
stochastic transitions between different conducting states of the nanowire can be induced , with a switching time as short as picoseconds . with an appropriate choice of dielectric
, the transconductance of the device can significantly exceed the conductance quantum @xmath1 , a remarkable figure of merit for a device at this lengthscale .
department of physics and astronomy , sacramento state university , 6000 j street , sacramento , ca 95819 department of physics , university of arizona , 1118 e. fourth street , tucson , az 85721 department of physics and courant institute of mathematical sciences , new york university , new york , ny 10003
|
in statistical mechanics , small systems weakly coupled to a large bath are described by canonical ensembles when the composite system ( system + bath ) is in a microcanonical state@xcite .
when we deal with a closed many - body quantum systems , the reduced density matrix of a small part of the system can be proven to be almost canonical , even if the state of the overall system is pure @xcite .
such gibbs - like behaviour emerges locally , in closed systems , as a direct consequence of the concentration of measure phenomenon .
its application to quantum statistical mechanics goes under the name of `` canonical typicality''@xcite .
it is a purely kinematic analysis on the hilbert space of the system and it can be proven in full generality by means of levy s lemma@xcite . a brief summary of canonical typicality and of levy s lemma can be found in appendix [ app : typ ] and [ app : levy ] . + in a recent paper@xcite
, the authors showed that typicality arguments can be fruitfully applied to the hilbert space of the spin - network states . in this context
, typicality provides a remarkable tool to study the local behaviour of a quantum geometry , in a fully kinematic approach , hence consistently with the general covariant nature of the spin - network description . in this letter
we propose a radical shift of setting where the role of `` system '' and `` bath '' is played by the boundary and bulk degrees of freedom of a generic _ ball _ of 3d quantum space .
we prove that , whenever the hilbert space of the boundary is much smaller than the bulk space , the reduced state of the boundary is always a thermal state , regardless what is the specific pure state of the whole region . in a series of recent works ,
such thermal states of the boundary have been interpreted as the pre - geometric equivalent of black hole configurations@xcite .
our general argument then comes in strong support to this vision , suggesting an information - theoretical origin for such proto - black - hole states .
+ there is more .
we find that the typical character of such thermal state of the boundary is regulated by a _ threshold condition _ which reproduces , at the pre - geometric level , the famous threshold mechanism of thorne s hoop conjecture @xcite .
we are then led to read this behaviour as an indirect proof of the hoop conjecture , in a purely quantum information - theoretic fashion .
in many background - independent approaches to quantum gravity , loop quantum gravity@xcite , spinfoams@xcite , group field theory@xcite , the microscopic quantum structure of space - time is described by purely combinatorial and algebraic variables , in terms of non - geometric structures realised by spin - networks@xcite .
they are generic graphs @xmath3 , made by vertices ( or nodes ) @xmath4 which are connected by edges @xmath5 .
the edges are coloured with irreducible representations of the local gauge group of the theory . in this case
the lorentz group , gauge - fixed to su(2 ) .
therefore to each @xmath6 one associates an @xmath1 irreducible representation ( irrep ) labelled by a half - integer @xmath7 called spin . the representation ( hilbert )
space is denoted @xmath8 and has dimension @xmath9 .
to each node of the graph @xmath4 one attaches an intertwiner @xmath10 .
this is an @xmath1-invariant map between the representation spaces @xmath8 associated to all the edges @xmath5 meeting at the vertex @xmath4 , @xmath11 in other words , @xmath10 is an invariant tensor , or a singlet state , between the representations attached to all the edges linked to the considered vertex .
once the @xmath12 s are fixed , the intertwiners at the vertex @xmath4 actually form a hilbert space @xmath13 $ ] .
a spin network state @xmath14 is defined then as the assignment of representation labels @xmath15 to each edge and the choice of a vector @xmath16 for the vertices .
therefore , for a given graph @xmath3 , upon choosing a basis of intertwiners for every assignment of representations @xmath17 , the spin networks provide a basis for the space of square - integrable wave - functional , endowed with the natural scalar product induced by the haar measure on @xmath1 @xmath18= \bigoplus_{\{j_e\ } } \bigotimes_v \mathcal{h}_v .\end{aligned}\ ] ] from a geometrical point of view , given a cellular decomposition of a three - dimensional manifold , a spin - network graph with a node in each cell and edges connecting nodes in neighbouring cells is said to be dual to this cellular decomposition .
therefore , each edge of the graph is dual to a surface patch intersecting the edge and the area of such patch is proportional to the representation @xmath12 .
analogously , vertices of a spin network can be dually thought of as chunks of volume ( see @xcite for the geometric interpretation of spin - networks states as collection of polyhedra )
. see figure ( [ fig : poly ] ) for an example . .
a node is dual to a chunk of space , with the shape of a polyhedron with a number of faces equal to the number of links around the node.,width=3 ] in particular , a bounded region @xmath0 , dual to a subregion of a generic spin network graph @xmath3 , includes a finite set of vertices ( @xmath19 ) and the edges connecting them ( @xmath20 ) .
we define a boundary @xmath2 as the set of edges which have only one end vertex laying in @xmath0 .
their number is called @xmath21 .
consistently , bulk edges are paths connecting vertices in @xmath0 .
we can picture @xmath0 as a 3-ball and @xmath2 as its boundary 2-sphere punctured by the boundary edges .
thanks to the gauge invariance at each node inside @xmath0 , we can gauge - fix this space and simplify the structure of bulk , without loosing any information @xcite .
remarkably , the gauge - invariant hilbert space associated the original graph @xmath22 is isomorphic to the gauge - invariant space defined on a different graph @xmath23 .
such graph consists of a single vertex intertwining the external edges of @xmath22 together with a certain number of loops @xmath24 which depends on the internal structure of @xmath22 .
the number of loops is @xmath25 , where @xmath26 and @xmath20 are , respectively , the number of nodes and edges inside @xmath22 . if @xmath27 there are no loops and @xmath22 has trivial topology .
such graph is usually called _ _ flower graph _ _ + such gauge reduction isomorphism does not produce any coarse graining , nevertheless the correspondence between the original graph and its flower is many - to - one , due to the fact that the procedure discards the combinatorial information about the internal subgraph .
an example is given in figure ( [ flowers])(a - c ) .
d ) graphic representation of the unfolding of the single vertex graph , with virtual link in red.,width=2 ] the hilbert space of @xmath23 provides us with the most synthetic description for a region of quantum space with non - trivial internal degrees of freedom .
in order to make our argument transparent we consider a simplifed set - up by working with graphs with fixed spin representations @xmath28 .
this means that the gauge - invariant hilbert space of our idealised 3-ball region is given by @xmath29\end{aligned}\ ] ] within @xmath23 we identify the _ boundary _ and _ bulk _ degrees of freedom of the 3-ball respectively with open edges and internal loops .
the separation is motivated by the different geometric information they carry .
loops are due to the presence of topological defects of the graph and carry localised `` curvature excitations '' at the vertices @xcite . on the other hand , the irreps carried by the open edges are dual to patches of the boundary surface . starting from such separation of degrees of freedom
, we can think of @xmath30 as a _
bipartite _ quantum mechanical system , with boundary and bulk subspaces weakly coupled by the constraint of @xmath1 gauge invariance at the vertex . in these terms our constrained space @xmath30 can be seen as embedded in a tensor product space @xmath31 where we define the ( unconstrained ) boundary and bulk hilbert spaces , respectively , as @xmath32 the coupling induced by the request of gauge invariance translates into correlations between boundary and bulk .
+ within @xmath30 , a consistent reorganisation of boundary and bulk degrees of freedom is realised by considering the unfolding of the single vertex flower graph in a graph with two vertices , recoupling edges and loops irreps , linked by a virtual edge .
an example is given in figure ( [ flowers])(d ) and it corresponds to the following re - writing of the gauge - invariant space @xmath33 where @xmath34 runs over the irreps of the virtual link .
moreover , @xmath35 and @xmath36 and @xmath37,@xmath38 counts the degeneracies of the space @xmath39 in the boundary and in the bulk recoupling .
the details of the way in which such decomposition is done can be found in the supplementary material , along with the information about the dimensions of @xmath40 and @xmath41 .
here we only need to fix the notation .
if we write a generic decomposition as @xmath42 , we call @xmath43 . all the dimensions needed in this paper can be written using @xmath44 .
more details can be found in the supplementary material .
as far as an observer _ external _ to the region @xmath0 is concerned , the geometry of the region @xmath0 is described by the information measured on the boundary of such region @xmath2 .
such information is given by the _ reduced _ density matrix obtained by taking the partial trace of the whole state @xmath45 over the non - trivial holonomies around the loops ( bulk degrees of freedom ) @xmath46 $ ] .
+ starting from the description of a region of quantum space given before , we focus on the state of the boundary . using the typicality tools ( summarised in appendix [ app : typ ] )
we prove that _ whenever the dimension of the bulk hilbert space exceeds the dimension of the boundary space , the reduced boundary state is always extremely close to the canonical state on the boundary @xmath47 , regardless what is the global state of the whole region_. the _ canonical state _ of the boundary @xmath47 is defined in the following way .
first we need @xmath48 , the microcanonical state over @xmath49 : @xmath50 where @xmath51 projects the states defined on @xmath52 onto the @xmath1 gauge invariant subspace @xmath49 , while @xmath53 is the trace of @xmath51 .
@xmath47 is the partial trace of @xmath54 over the bulk degrees of freedom : @xmath55 the quantity which we are interested in is the trace - distance@xcite between a generic reduced state @xmath56 and @xmath47 : @xmath57 . using the prescription developed in @xcite and summarised in appendix [ app : typ ] , it is possible to show that its average over the global hilbert space @xmath58 $ ] satisfies @xmath59 \leq \frac{d_{\partial r}}{\sqrt{d_{\mathcal{f}_r } } } \end{aligned}\ ]
] moreover , it can be proven that the fraction of states which are @xmath60 away from this average is incredibly small @xmath61 } { \mathrm{vol } \left [ \phi \in \mathcal{h}_{\mathcal{f}_r } \right ] } \leq \eta ' \label{result}\end{aligned}\ ] ] where @xmath62 & & \eta < \varepsilon + \frac{1}{2 } \frac{d_{\partial r}}{\sqrt{d_{\mathcal{f}_r } } } , \end{aligned}\ ] ] therefore , whenever the right - hand side of eq.([eq : bound ] ) is much smaller than one , it will be concretely impossible to distinguish the actual reduced state @xmath56 from @xmath47 .
the goal is then to evaluate this bound and find the regime where the distance is effectively pushed to zero .
building on the technical results derived in @xcite , we work in the regime @xmath63 , which goes along with our statistical approach . using the expression derived in @xcite
we obtain @xmath64^{3/4}\end{aligned}\ ] ] the details can be found in the supplementary material .
this quantity has a leading exponential behaviour , both in the number of external edges and in the number of loops over which we trace .
the exponent becomes negative as soon as @xmath65 .
such fast decay is present for any choice of @xmath28 , even far from the `` semiclassical regime '' @xmath66 .
for example in the smallest case @xmath67 we have @xmath68^{3/4}\end{aligned}\ ] ] this is not exactly a threshold behaviour but it is a fast exponential decay , which becomes faster as we approach the semiclassical regime @xmath66 . in that regime the exponential decay of
@xmath69 $ ] to zero approaches precisely a threshold behaviour , regulated by the condition @xmath70 will will discuss the physical meaning of this condition in the final section .
for the time being , we note that the left - hand side is proportional to the total area while the right - hand side is proportional to the curvature excitations carried by the internal loops . intuitively speaking
, since the trace of the loop holonomy is the curvature around a path , in the semiclassical limit an increase in the number of internal loops corresponds to an increase of the gravitational energy _ density _ within the bounded @xmath0 region dual to @xmath22 .
therefore the condition can be loosely interpreted as an inequality relating the area of a closed surface with the gravitational energy density inside .
this suggests a connection with the inequality in the hoop conjecture .
we now focus on the explicit form of the typical state of the boundary . starting from the decomposition of the constrained space given in eq.([unfolded2 ] ) , a convenient basis in either of the two subspaces is labeled by three numbers , respectively @xmath71 and @xmath72 , with @xmath73 running over the degeneracy of the irrep @xmath39 at fixed value of @xmath34@xcite .
where @xmath147 $ ] stands for `` the volume of states @xmath145 such that @xmath148 '' .
@xmath149 is the average of the function @xmath143 over the whole hilbert space and @xmath150 $ ] is the total volume of the hilbert space .
integrals over the hilbert space are performed using the unique unitarily invariant haar measure .
+ the levy lemma is essentially needed to conclude that all but an exponentially small fraction of all states are quite close to the canonical state .
this is a very specific manifestation of a general phenomenon called `` concentration of measure '' , which occurs in high - dimensional statistical spaces @xcite .
the effect of such result is that we can re - think about the `` a priori equal probability '' principle as an `` apparently equal probability '' stating that : as far as a small system is concerned almost every state of the universe seems similar to its average state , which is the maximally mixed state @xmath151 .
+ w. greiner , l. neise , h. stocker , _ thermodynamics and statistical mechanics _ ( springer , 1995 ) j. uffink - _ compendium of the foundations of classical statistical physics _ , chapter for `` handbook for philosophy of physic '' , j. butterfield and j. earman ( eds ) ( 2006 ) e. schroedinger - _ statistical thermodynamics _ ,
( dover publications , 1989 ) s. popescu , a. j. short , and a. winter , _ entanglement and the foundations of statistical mechanics _
, nature physics 2 , 754 ( 2006 ) ; charles , c. & livine , e.r . _ the fock space of loopy spin networks for quantum gravity _ , gen .
* 48 * : 113 ( 2016 ) livine e r and terno d r _ quantum black holes : entropy and entanglement on the horizon _
nucl.phys.b 741 131 ( 2006 ) livine e r and terno d r _ bulk entropy in loop quantum gravity _
nucl.phys.b 794138 ( 2008 ) k. thorne , _ nonspherical gravitational collapse : a short review _ , in magic without magic : john archibald wheeler , ed .
j. klauder ( w.h .
freeman & co. , san francisco , 1972 ) .
e. flanagan , _ hoop conjecture for black - hole horizon formation _ , phys .
d * 44 * , 2409 ( 1991 ) f. pastawski , b. yoshida , d. harlow , and j. preskill , _ holographic quantum error - correcting codes : toy models for the bulk / boundary correspondence _ , journal of high energy physics 149 ( 2015 ) [ arxiv:1503.06237 ] .
qi , _ exact holographic mapping and emergent space - time geometry _ , arxiv:1309.6282
. s. ryu , t. takayanagi , _ aspects of holographic entanglement entropy _ , jhep 0608:045 , arxiv : hep - th/0605073 , ( 2006 ) .
o. t. courtney , g. bianconi , _ generalized network structures : the configuration model and the canonical ensemble of simplicial complexes _
e 93 , 062311 ( 2016 ) , arxiv:1602.04110 [ physics.soc-ph ]
|
we consider a closed region @xmath0 of 3d quantum space modeled by @xmath1 spin - networks . using the concentration of measure phenomenon
we prove that , whenever the ratio between the boundary @xmath2 and the bulk edges of the graph overcomes a finite threshold , the state of the boundary is always thermal , with an entropy proportional to its area .
the emergence of a thermal state of the boundary can be traced back to a large amount of entanglement between boundary and bulk degrees of freedom . using the dual geometric interpretation provided by loop quantum gravity , we interprete such phenomenon as a pre - geometric analogue of thorne s `` hoop conjecture '' , at the core of the formation of a horizon in general relativity .
|
the rapid expansion of the internet in the last two decades has produced a large - scale system of thousands of diverse , independently managed networks that collectively provide global connectivity across a wide spectrum of geopolitical environments . from 1997 to 2005
the number of globally routable as identifiers has increased from less than 2,000 to more than 20,000 , exerting significant pressure on interdomain routing as well as other functional and structural parts of the internet .
this impressive growth has resulted in a heterogenous and highly complex system that challenges accurate and realistic modeling of the internet infrastructure . in particular
, the as - level topology is an intermix of networks owned and operated by many different organizations , e.g. , backbone providers , regional providers , access providers , universities and private companies .
statistical information that faithfully characterizes different as types is on the critical path toward understanding the structure of the internet , as well as for modeling its topology and growth . in topology modeling ,
knowledge of as types is mandatory for augmenting synthetically constructed or measured as topologies with realistic intra - as and inter - as router - level topologies .
for example , we expect the network of a dual - homed university to be drastically different from that of a dual - homed small company
. the university will likely contain dozens of internal routers , thousands of hosts , and many other network elements ( switches , servers , firewalls ) . on the other hand
, the small company will most probably have a single router and a simple network topology .
since there is such a diversity among different network types , we can not accurately augment the as - level topology with appropriate router - level topologies if we can not characterize the composing ases .
moreover , annotating the ases in the as topology with their types is a prerequisite for modeling the evolution of the internet , since different types of ases exhibit different growth patterns .
for example , internet service providers ( isp ) grow by attracting new customers and by engaging in business agreements with other isps . on the other hand , small companies that connect to the internet through one or few isps
do not grow significantly over time .
thus , categorizing different types of ases in the internet is necessary to identify network evolution patterns and develop accurate evolution models .
an as taxonomy is also necessary for mapping ip addresses to different types of users .
for example , in traffic analysis studies its often required to distinguish between packets that come from home and business users .
given an as taxonomy , its possible to realize this goal by checking the type of as that originates the prefix in which an ip address lies .
in this work , we introduce a radically new approach based on machine learning to construct a representative as taxonomy .
we develop an algorithm to classify ases based on empirically observed differences between as characteristics .
we use a large set of data from the internet routing registries ( irr ) @xcite and from routeviews @xcite to identify intrinsic differences between ases of different types .
then , we employ a novel machine learning technique to build a classification algorithm that exploits these differences to classify ases into six representative classes that reflect ases with different network properties and infrastructures .
we derive macroscopic statistics on the different types of ases in the internet and validate our results using a sample of 1200 manually identified as types .
our validation demonstrates that our classification algorithm achieves high accuracy : 78.1% of the examined classifications were correct .
finally , we make our results and our classifier publicly available to promote further research and understanding of the internet s structure and evolution . in section [ sec : rel ] we start with a brief discussion of related work .
section [ sec : data ] describes the data we used , and in section [ sec : classes ] we specify the set of as classes we use in our experiments .
section [ sec : clas ] introduces our classification approach and results .
we validate them in section [ sec : val ] and conclude in section [ sec : con ] .
several works have developed techniques decomposing the as topology into different levels or tiers based on connectivity properties of bgp - derived as graphs .
govindan and reddy @xcite propose a classification of ases into four levels based on their as degree .
et al_. @xcite classify ases into seven tiers based on inferred customer - to - provider relationships .
their classification exploits the idea that provider ases should be in higher tiers than their customers .
et al_. @xcite classify ases into five tiers based on inferred customer - to - provider as well as peer - to - peer relationships .
our work differs from previous approaches in the following ways : 1 .
we do not employ heuristics and ad - hoc thresholds to define the as levels .
instead , we use a novel machine learning algorithm to identify intrinsic features distinguishing different as types .
2 . we do not rely exclusively on as graphs , which often miss a substantial fraction of the true as links in the internet , resulting in incomplete as topologies .
instead , we use an extensive set of diverse data including irr records , inferred as relationships , ip prefixes , and as graphs
we do not classify ases into hierarchies of levels or tiers extracted from as graphs using degree - based or more sophisticated mechanisms ; these methods tend to mix ases with substantially different network properties into a single as group . instead
, we specify a representative set of as classes characterized by unique signatures of network properties .
to construct the set of as attributes that we use in our as classification , we collect data from the following databases and measurement projects : 1 ) * irrs * @xcite .
the irrs constitute a distributed database containing records on ases routing policies , assigned ip prefixes , contact information , etc . a natural approach to identifying the type of an as , given its as number ,
is to lookup the as number in the irrs and examine its organization description record . in the rpsl
@xcite terminology , this record is the descr attribute of the rpsl class aut - num .
it contains the name or a short description of the organization that owns the as number .
for example , the following are entries for the descr attribute found in the irrs : `` intervivo networks , a broadband internet access provider '' and `` auckland peering exchange ''
. the descr attribute does not have a standard representation .
it usually consists of a short description as in the examples above , but in some cases it only contains an acronym , e.g. , `` kpmg llp '' , `` lti '' . for the purposes of this work ,
the descr record is our first as attribute , from which we extract useful information by means of text classification techniques .
we downloaded the mapping of as numbers to organization description records on 04/08/2005 from the cidr report @xcite , which provides on a daily basis mappings of as numbers to organization description records extracted from arin , ripe ncc , lapnic , apnic , krnic , twnic , and jpnic databases .
we preprocess the organization records by removing stop words , i.e. , words with little semantic meaning , such as `` of '' or `` the '' , using the stop word list @xcite . then ,
using the porter stemming algorithm @xcite , we replace words with their stem .
we note that irrs contain significant portions of incomplete or obsolete records , which is not a serious problem for this study since we are only concerned with the descr attribute , which changes relatively rarely .
2 ) * routeviews * @xcite .
routeviews is a measurement project that collects and archives a union of bgp tables from a large number of isps .
we download all 12 bgp table snapshots archived from the collector route-views2.oregon-ix.net on 07/18/2005 . for each table snapshot
we extract as paths and remove as sets and private as numbers .
then , we merge the extracted as paths into an as topology and use the as relationship inference heuristics of @xcite to annotate the as links with customer - to - provider and peer - to - peer relationships . having the as relationships inferred , we calculate the following three attributes for every as : the number of providers , the number of customers , and the number of peers a given as is connected to .
large isps typically have a large number of customers , zero providers , and a small number of peers , while small isps typically have few providers , a small number of customers and a large number of peers .
stub university or company networks typically have zero customers , zero or few peers and a small number of providers . from the routeviews data , we also extract information on ip prefixes .
we use the chronologically first table snapshot from the same snapshot set to construct a mapping of as numbers to the ip prefixes they advertise .
then , for each as we count the number of advertised prefixes and use this number as another as attribute .
small ases , with tiny portions of ip address space allocated to them , as well as older ases with large ip blocks , tend to advertise few prefixes . on the contrary , large ases with relatively high ip address utilization and diversified routing policies tend to advertise many prefixes of various lengths .
one problem with this attribute is that ip prefixes are of drastically different sizes .
advertised ipv4 prefixes range in size from a /8 , covering @xmath0 ip addresses , down to a /32 , covering just one address .
the prefix length of 24 bits is generally the smallest ipv4 prefix size that is globally routed , which suggests our last as attribute to be the number of unique /24 prefixes found within the union of address space advertised by the as .
this attribute is likely to have small values for smaller ases that use and advertise smaller portions of ip address space , while it is likely to be at its maximum for large or old ases ( those that appeared in the internet early , e.g. , some military and academic networks ) since they often have huge chunks of assigned ip address space that they utilize scarcely . in summary , we collect data from the irrs and routeviews .
we find 19,537 ases .
using the collected data , we annotate every as with the following six attributes : 1 ) the organization description record ( the _ description _ attribute ) , 2 ) the number of inferred customers ( the _ customer _ attribute ) , 3 ) the number of inferred providers ( the _ provider _ attribute ) , 4 ) the number of inferred peers ( the _ peer _ attribute ) , 5 ) the number of advertised ip prefixes ( the _ prefix _ attribute ) , and 6 ) the equivalent number of /24 prefixes covering all the advertised ip space ( the _ space _ attribute ) .
in this work , we focus on network properties of an as as the main criterion determining the set of as classes . in other words , to construct the as class set , we use the rule that ases in the same class should have similar network properties , while ases in different classes should have different network properties .
ases in the same class may still have significant network differences , however these differences should be small compared with differences between networks of different classes . for example , a small university and a large university may have quite different networks , however these differences are less significant compared to the differences between a typical university network and a typical isp network .
besides employing our _ de facto _ empirical knowledge , we perform the following experiment to specify the set of as classes .
we _ randomly _ select 1200 ases and then , for each as , we examine its attributes , visit its website ( if possible ) , search for references to its organization name and determine its business profile .
after examining the spectrum of these 1200 ases , we construct our as class set : 1 .
_ large isps _ : large backbone providers , tier-1 isps , with intercontinental networks . 2 .
_ small isps _ : regional and access providers with small metropolitan or larger regional networks .
_ customer ases _ : companies or organizations that run their own networks but as opposed to members of the previous two classes do not provide internet connectivity services .
we find a wide range of ases in this class , like web hosting companies , technology companies , consulting companies , hospitals , banks , military networks , government networks , etc .
universities _ : university or college networks .
we distinguish these networks from members of the customer as class , since they typically have substantially larger networks that serve thousands of end hosts .
5 . _ internet exchange points ( ixps ) _ : small networks serving as interconnection points for the members of the first two classes . 6 . _ network information centers ( nics ) _ : networks that host important network infrastructure , such as root or tld servers .
we build our classification algorithm using _ adaboost _
@xcite , a very powerful machine learning technique .
the main idea behind adaboost is to combine multiple simple classification rules into a efficient composite classifier .
these simple classification rules are called _ weak hypotheses _ and by definition are only required to perform slightly better than random guessing .
intuitively , weak hypotheses reflect simple `` rules of thumb '' that are usually much easier to construct than a complex classifier .
adaboost works iteratively over a set of training examples ; at each iteration it finds a weak hypothesis that performs well on the examples which the weak hypotheses of previous iterations erroneously classified .
one constructs a weak hypothesis by means of a _ weak learning algorithm _ or simply _
weak learner_. the power of adaboost lies in a well - developed theoretical framework that intelligently combines the weak hypotheses into a composite classifier .
let @xmath1 denote the set of ases that we want to classify and @xmath2 be the set of possible classes , such that each as @xmath3 belongs to a unique class in @xmath2 . if @xmath4 is the correct class for as @xmath5 , than let @xmath6 , otherwise @xmath7 .
the goal is to find a classifier that for each as produces a
_ ranking _ of all the possible classes .
more formally , we will compute a ranking function @xmath8 : for each as @xmath3 , the classes in @xmath2 will be ordered according to @xmath9the higher the value of @xmath10 , the more likely @xmath5 belongs to @xmath11 . in figure
[ fig : ada ] , we illustrate the adaboost.mh algorithm @xcite , a special member of the adaboost algorithm family that is suited for solving multiclass problems . let @xmath12 be the set of training examples .
in our case , we construct @xmath13 by manually determining the correct class of 1220 ases : 1200 are randomly selected ; and 20 are well - known large isps and ixps , which we use to increase the number of these two types of ases in the initial random sample .
let @xmath14 be the total number of iterations .
for each iteration @xmath15 , we maintain a distribution @xmath16 of weights over the set of examples and classes @xmath17 . at the first iteration , we initialize @xmath16 to the uniform distribution , i.e.
, @xmath18 is constant for all @xmath19 , @xmath20 . at each subsequent iteration , we pass the distribution @xmath16 and the training examples @xmath13 to a weak learner that computes a weak hypothesis @xmath21 .
a positive ( negative ) sign of the weak hypothesis @xmath22 corresponds to the prediction that as @xmath5 is ( is not ) a member of class @xmath11 .
the value @xmath23 of the weak hypothesis reflects the confidence level of the prediction .
then , we update the distribution @xmath16 so that the @xmath5 , @xmath11 pairs that were erroneously predicted , i.e. , the signs of @xmath24 and @xmath25 differ , receive a exponentially higher weight . by assigning higher weight to incorrect predictions , we force the algorithm to focus on these difficult examples in the next round .
the final classifier @xmath26 is the sum of votes of the weak hypotheses in all rounds @xmath27 , @xmath15 .
a weak hypothesis is equivalent to a one - level decision tree that checks a single as attribute . for the description as attribute , a weak hypothesis searches for the presence of a term or a sequence of terms in a given record , and
if a match occurs , it outputs a confidence value for each of the classes .
for example , upon finding the term `` university '' in the record `` seoul national university of education '' the weak hypothesis will likely output a high positive confidence value for the university as class and a negative confidence value for the other as classes . for scalar attributes , a weak hypothesis asks if a given attribute value is above or below a certain threshold .
depending on the outcome , the hypothesis outputs different confidence values .
the weak learner builds a weak hypothesis by exhaustively evaluating the attributes in the given weighted training examples . for a text attribute , it builds a candidate weak hypothesis by evaluating all possible terms and sequences of terms . for each term or sequence of terms
, it calculates the appropriate confidence values by minimizing the hamming loss , which is the fraction of examples @xmath5 and classes @xmath11 , for which the sign of the final classifier @xmath10 differs from @xmath25 .
similarly , for each scalar attribute , the weak learner builds a candidate weak hypothesis by exhaustively searching the threshold and confidence values minimizing the hamming loss . on its output
, the learner returns the weak hypothesis attaining the minimum hamming loss .
ccc|cccccc round & attribute & term / threshold & l.isp & s.isp & cusmr & uni & ixp & nic + & & @xmath28 & ( 1,7.2 ) ( 1,-0.0163121304999998)(1,3.5163121305 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.4735347919)(1,1.0264652081 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.7797397768 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.8325260019)(1,1.6674739981 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.4639146476)(1,2.0360853524 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.2714864672)(1,2.2285135328 ) ( 0,3.5)(1,0)3 + & & @xmath29 & ( 1,7.2 ) ( 1,1.2708921135)(1,2.2291078865 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.0336004254 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.1608508751)(1,0.3391491249 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.2256887679)(1,1.2743112321 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.5301809138)(1,1.9698190862 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.990387677)(1,1.509612323 ) ( 0,3.5)(1,0)3 + 2 & description & `` network inform '' & ( 1,7.2 ) ( 1,2.2216833816)(1,1.2783166184 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.1210242844)(1,2.3789757156 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.1349191402)(1,2.3650808598 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.7399693498)(1,1.7600306502 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.0556227578)(1,1.4443772422 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,3.2284513535 ) ( 0,3.5)(1,0)3 + & & @xmath30 & ( 1,7.2 ) ( 1,0.6010428496)(1,2.8989571504 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.4404276559)(1,0.0595723441 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.1111626141 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.0390868724 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.4052849295)(1,0.0947150705 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.7965965279)(1,0.7034034721 ) ( 0,3.5)(1,0)3 + & & @xmath31 & ( 1,7.2 ) ( 1,3.5)(1,0.8634883271 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.7049122965 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.3929064144)(1,1.1070935856 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.1135206426)(1,0.3864793574 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.4135959469 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.7023153932)(1,0.7976846068 ) ( 0,3.5)(1,0)3 + 4 & description & `` exchang '' & ( 1,7.2 ) ( 1,2.6649888658)(1,0.8350111342 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.2936841338)(1,2.2063158662 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.6048529329)(1,0.8951470671 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.9378176669)(1,1.5621823331 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,2.4512993094 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.4186244687)(1,1.0813755313 ) ( 0,3.5)(1,0)3 + 5 & description & `` univers '' & ( 1,7.2 ) ( 1,2.7987866841)(1,0.7012133159 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.8342091922)(1,1.6657908078 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.0500111654)(1,1.4499888346 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,2.4096956442 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.1049456495)(1,1.3950543505 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.263214415)(1,1.236785585 ) ( 0,3.5)(1,0)3 + & & @xmath32 & ( 1,7.2 ) ( 1,0.9666483892)(1,2.5333516108 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.0514789400 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.0423646810 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.3302953706)(1,0.1697046294 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.9781279276)(1,0.5218720724 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,3.5)(1,0.0133931961 ) ( 0,3.5)(1,0)3 + & & @xmath33 & ( 1,7.2 ) ( 1,3.5)(1,2.4854406092 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,1.3250967327)(1,2.1749032673 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.3689860715)(1,1.1310139285 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.4639524202)(1,1.0360475798 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.4179592278)(1,1.0820407722 ) ( 0,3.5)(1,0)3 & ( 1,7.2 ) ( 1,2.7330272347)(1,0.7669727653 ) ( 0,3.5)(1,0)3 + to realize our classification algorithm , we use boostexter @xcite , a publicly available implementation of adaboost . in table [ table ] we depict the weak hypotheses that our algorithm discovered during its first six iterations . for each weak hypothesis , we illustrate the selected as attribute , the term or threshold that is looked for , and the computed confidence values .
the first weak hypothesis deals with the space attribute . if the ip address space advertised by an as is less than 8.5 ( equivalent ) /24
prefixes then , the hypothesis assigns a positive confidence value to the customer as class and negative confidence values to the other classes .
if the value of the space attribute is above 8.5 , the hypothesis assigns negative or very close to zero confidence values to all the classes , which means that in this case it can not make a confident positive prediction .
the second weak hypothesis checks the description as attribute for the presence of term `` network inform . ''
if it finds one , it assigns a high positive confidence to the nic as class and negative confidence values to the other classes .
note , that in some cases the weak hypothesis assigns zero confidence values , meaning that it abstains from making any prediction .
we experimentally fix the number of rounds @xmath14 to 28 , since subsequent iterations lead to overfitting .
overfitting is a common problem in machine learning .
it is a consequence of too extensive training of an algorithm on one dataset .
the undesired effect is that the algorithm is memorizing the training examples instead of extracting concepts from them .
fortunately , we can easily detect if the algorithm tends to overfit by examining the produced classification rules . having the number of iterations fixed
, we finally apply our classifier to the set of 19,537 ases and calculate the ranking of the as classes for them . for each as we consider the highest ranked class as its predicted class . if the class with the highest rank value for an as has the confidence value less than or equal to zero , then the classifier abstains from making a prediction since the given information is not sufficient to produce a reliable assignment .
overall , the classifier abstains from making a prediction for 923 ases , which accounts for 4.7% of the total number of ases in our dataset . in table [ stats ] we show the per category classification statistics . among the classified ases , 63.0% are customers , 30.1% are small isps , 4.7% are universities , 1.8% are nics , 0.2% are isps , and 0.2% are large ixps .
.numbers of ases in each as class . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ]
to validate our results , we employ the standard machine learning methodology called _ cross - validation_. cross - validation is the process of splitting the training examples into two subsets .
one then uses the first subset to train a new classifier and the second subset to validate the results of this new classifier . from our main set of 1200 training examples ,
we randomly extract 1100 ases and use varying size subsets of these 1100 ases to train new classifiers .
we validate the predictions of the new classifiers against the remaining 100 examples .
we repeat the random selection process 400 times and for each iteration we compute the following evaluation metrics : 1 ) _ accuracy _ , which we define as the percentage of ases for which the as class with the highest rank value is their correct class .
the disadvantage of this metric is that it checks only the top of the ranking , ignoring the remaining positions .
to address this problem , we use 2 ) _ coverage _ , which we define as the average position number of the correct class of an as . for each as ,
we number as class positions incrementally starting from zero for the class with the highest positive confidence value .
thus , if all the predictions are correct the coverage is zero . in figures [
plot : accuracy ] and [ plot : coverage ] we plot the average accuracy and coverage versus the size of the training set . as the size of the training set increases , the accuracy increases and the coverage decreases . for @xmath34 the accuracy reaches 0.781 , e.g. , 78.1% , and the coverage 0.251 .
the increasing accuracy trend suggests that for the training size of 1200 that we use in our final classification , the expected accuracy must be even higher .
the low value of the coverage indicates that when the correct class is not of the top rank value , it is close to the top .
more specifically , we find that for 97.7% of the predictions the correct class is in the top two positions of the ranking .
we next analyze the per class percentage of correct predictions .
we find that for @xmath34 the percentage of correct predictions is on average : 100% for large isps , 100% for nics , 100% for ixps , 92.8% for universities , 79.2% for customer ases and 72.1% for small isps .
the actual distribution of ases among the classes in our training set is : 684 customer ases , 401 small isps , 66 universities , 36 nics , 11 ixps , and 2 large isps .
the lower accuracy for customer ases and small isps illustrates that these are the hardest classes to identify .
we explain this effect by similarities between the characteristics of these two as classes : 1 ) more than a half of the small isps appear to have the as degree of 1 or 2 , which is also typical for customer ases ; 2 ) some customer ases , especially web hosting companies , advertise large numbers of different ip prefixes or large chunks of address space , which is also typical for isps . in summary , we find that in the examined examples our classifier almost perfectly identifies large isps , nics , ixps and universities , while it also produces accurate predictions for customer ases and small isps , which are the hardest to classify .
in this work , we establish a classification of ases required for expanding our understanding of the internet infrastructure and for creating realistic models of its topology and evolution .
we develop a novel classification methodology that we apply to an exhaustive set of as data to obtain the first statistics on the different as classes in the internet .
we validate our results and demonstrate that our classifier achieves accuracy of 78.1% in the examined data . to promote further analysis and to inspire development of better topology models , we release to the community our classification dataset along with the as class predictions @xcite . to the best of our knowledge ,
our dataset is the most comprehensive collection of as macroscopic characteristics .
in addition to as topology and taxonomy information , it includes organization description records , as business relationship information , and information on advertised ip prefixes and space .
|
although the internet as - level topology has been extensively studied over the past few years , little is known about the details of the as taxonomy .
an as `` node '' can represent a wide variety of organizations , e.g. , large isp , or small private business , university , with vastly different network characteristics , external connectivity patterns , network growth tendencies , and other properties that we can hardly neglect while working on veracious internet representations in simulation environments . in this paper
, we introduce a radically new approach based on machine learning techniques to map all the ases in the internet into a natural as taxonomy .
we successfully classify 95.3% of ases with expected accuracy of 78.1% .
we release to the community the as - level topology dataset augmented with : 1 ) the as taxonomy information and 2 ) the set of as attributes we used to classify ases .
we believe that this dataset will serve as an invaluable addition to further understanding of the structure and evolution of the internet .
|
ob runaways are a subgroup of spectral type o and b stars which are recognized by their unusually high velocity ( @xmath7 ) , often away from known star forming regions ( blaauw , 1961 ) . stone ( 1979 ) also included stars at large distance ( up to several kpc ) from the galactic plane for which no parent association could be found .
about 40 per cent of o stars and 5 - 10 per cent of b stars are runaways ( stone 1991 ) , and only one in ten ob runaways has a known binary companion ( gies & bolton 1986 ) .
the peculiar velocities of about @xmath8 ob runaways have been determined and the trajectories of @xmath920 can be traced back to a nearby association ( hoogerwerf et al . 2000 , 2001 ) , supporting the idea that these stars originated in stellar clusters from which they were later ejected with high velocities .
the two main mechanisms that have been proposed to explain the high velocities of ob runaways are : * ejection upon a supernova in a binary system ( zwicky 1957 ; blaauw 1961 ) ; * ejection by a close multiple encounter in a star cluster ( poveda , ruiz & allen , 1967 ) .
we refer to scenario 1 and 2 as _ supernova ejection _ and _ dynamical ejection _ , respectively . in the supernova ejection scenario
the ob runaway gets a high velocity from the explosion of the companion star .
if the binary is dissociated by the supernova , the companion star is ejected with a velocity of the order of its orbital velocity while if the binary remains bound , the mass loss in the supernova induces a recoil velocity in the center of mass of the binary ( blaauw 1961 ; hills 1980 ; tauris & takens 1998 ) .
this scenario predicts that about 30 per cent of the ob runaways should still be in a binary with a compact companion .
large natal kicks ( up to 1000@xmath3 ) are observed in isolated pulsars ( lyne & lorimer 1994 ; cordes & chernoff 1998 ) , but are not sufficient to explain the low binarity fraction observed for ob runaways .
even if the formation of a neutron star is accompanied by a kick drawn from a maxwellian distribution with a mean of 600@xmath3 , about 30 per cent of the binaries remain bound ( portegies zwart 2000 ) .
selection effects may be important in preventing the observation of compact companions to the runaways ( e.g. colin et al .
1996 ) , but the supernova ejection scenario seems insufficient to explain all the known ob runaways ( portegies zwart 2000 ) . the supernova ejection scenario , however , satisfactorily explains the high velocities ( @xmath10 ) of some known high - mass x - ray binaries ( van den heuvel et al .
2000 ) . in one of these , vela x-1
, the high space velocity is inferred from the morphology of a bow shock ( kaper et al .
1997 ) while the neutron star is visible as an x - ray pulsar .
the only known example of a binary dissociated by a natal kick is given by the o9.5v star @xmath11 oph and the pulsar psr j1932 + 1059 .
the large rotational velocity ( penny 1996 ) and helium abundance ( herrero et al . 1992 ) of @xmath11 oph indicate that a phase of mass transfer occurred , as expected for runaways ejected by a supernova explosion . under this hypothesis ,
a natal kick of about 400@xmath3 can be inferred for the pulsar ( hoogerwerf , de bruijne & de zeeuw 2001 ) .
an alternative scenario to the supernova ejection is the dynamical ejection scenario ( poveda , ruiz & allen 1967 ; gies & bolton 1986 ; leonard 1990 ) , in which close encounters between binaries and/or single stars in dense stellar regions can lead to the ejection of stars with high velocity .
young stellar clusters and ob associations in early evolutionary stages can have high stellar densities ( @xmath12 ) in their cores and therefore close encounters between single stars , binaries or higher order systems can be very frequent .
these encounters can result in the ejection of stars with high velocities .
o and b stars are more likely to participate in strong encounters than low mass stars because of mass segregation and gravitational focusing . as a consequence ,
the number of close encounters involving o and b type stars can be high despite their short lifetime , especially when the cluster is young and the binary fraction is high . to help understanding the runaway mechanism we define the _ kinematical age _
@xmath13 as the time since the ejection from the parent association .
this time can be derived integrating the trajectory of the runaway in the galactic potential . at first order @xmath14 , where @xmath15 is the distance traveled from the cluster and @xmath16 the velocity of the runaway .
the kinematical age can help to identify the ejection mechanism .
a runaway which is ejected by a supernova explosion has a kinematical age smaller than the age of the parent cluster because the primary star in the original binary evolved for a few million years before it exploded ejecting its companion .
dynamical ejection is likely to occur when the cluster is very young and therefore in this case the kinematical age is often similar to the age of the parent system .
an example of stars ejected by a dynamical encounter is provided by the single runaways @xmath0 and @xmath1 and the spectroscopic binary @xmath2 , whose trajectories in the galactic potential have been traced back in time and intersect about @xmath5myr ago at the expected location of the trapezium cluster ( hoogerwerf et al .
the runaways @xmath0 and @xmath1 show no evidence of binary evolution , like a high rotational velocity or an increased helium abundance , and move in almost opposite directions away from the trapezium cluster .
the kinematical and evolutionary properties of the system suggest the hypothesis that the stars were involved in a 4-body encounter that ejected them with high velocity . in this paper
we describe a combined use of direct n - body simulations and stellar and binary evolution calculations to identify the type of encounter that ejected @xmath0 , @xmath1 and @xmath2 . the paper is organized as follows : in section @xmath17 [ sec : prop ]
we describe the main observational properties of the stars and a possible evolutionary model for @xmath2 , @xmath0 and @xmath1 and we propose a binary - binary encounter able to reproduce the current positions , velocities and physical properties of the four stars ; in section @xmath17 [ sec : binbin ] we describe the code used to perform simulations of 4-body encounters and the classification of the outcomes ; in section @xmath17 [ sec : simulations ] we describe the choice of initial conditions for the simulations , we present the main results obtained from the scattering experiments and we compare them with the observed properties of the stars while in section @xmath17 [ sec : evolution ] we present aspects of the evolution of the @xmath2 binary after the encounter that can affect the model for the dynamical encounter .
the single runaway stars @xmath20 and @xmath21 are o type stars moving away from the orion star forming region . blaauw & morgan ( 1954 ) recognized them as a remarkable pair of runaways having similar spectral types and moving in almost opposite directions with space velocities of about 100@xmath3 . at a distance of about @xmath22pc from the trapezium cluster
the kinematical age of both stars is about @xmath5myr .
the same kinematical age and parent cluster for the two stars together with their relative straight angle of motion suggest a common origin in a dynamical encounter .
gies & bolton ( 1986 ) advanced the hypothesis of a binary - binary interaction that ejected @xmath0 and @xmath1 and left @xmath2 as the surviving binary .
they noticed that @xmath23 is one of the most massive objects in the orion nebula , has a high eccentric orbit and its component stars have relative orbital velocities similar to the space velocities of @xmath20 and @xmath21 .
parallaxes and proper motions provided by _
hipparcos _ together with radial velocity measurements ( turon et al .
1992 ) are shown in table [ tab : vel ] and allow precise determinations of the galactic trajectories of the binary and the single stars .
the integration of the trajectories of @xmath20 , @xmath21 and @xmath23 ( hoogerwerf et al .
2001 ) in the galactic potential shows that @xmath24myr ago all the stars had a minimum separation of about 4@xmath25pc .
the position and velocity of the parent cluster , derived from the orbit integration and the 4-body center of mass velocity , are consistent with those of the trapezium cluster in the orion nebula . .parallax ,
distance from the sun and proper motions from the hipparcos catalogue ( esa 1997 ) , radial velocities from the hipparcos input catalogue ( turon et al .
1992 ) , velocity relative to the local standard of rest and velocity relative to the center of mass of the 4-body system for @xmath20 , @xmath21 and @xmath23 . [ cols="<,^,^,^",options="header " , ] in this section we relax the hypothesis of point mass stars and investigate the effects of physical radii on the outcomes of binary - binary encounters . in the simulations described in the previous sections , we keep track of the minimum distances between all stars . in this way we are able to use the same data and test the effect of finite stellar radii , assuming that stars collide if they approach each other within some minimum distance @xmath26 .
since the initial eccentricities of both binaries are randomly drawn from a thermal distribution between 0 and 1 , stars in very eccentric orbits may collide at the first pericenter passage .
this would not be realistic , and therefore we limit the range in initial eccentricities . in a highly eccentric orbit two stars with total radius @xmath27 collide if @xmath28 ( kopal 1959 ) . inverting this equation leads to our definition of @xmath29 .
for the observed stellar radii ( see table [ tab : scheme ] ) @xmath30 .
we therefore exclude binaries with initial eccentricity @xmath31 from further consideration . , between stars during binary - binary encounters .
the solid line represents the result for systems with @xmath32 from our standard run .
the dashed line is for binaries with initially circular orbits . with the observed stellar radii
a collision would occur at about @xmath33.,width=264 ] figure [ fig : cumulative ] shows the cumulative distribution for minimum distances between any two stars for the @xmath34 encounters which remained after selecting the systems with @xmath35 ( solid curve ) and for circular initial binaries ( dashed curve ) .
since circular orbits are very rare in our standard initial conditions , we performed a separate simulation of 8000 encounters .
if we assume stellar radii from tab.[tab : scheme ] at the moment of the encounter , a collision occurs if @xmath33 . in that case
about half of the initially circular binaries result in a collision , whereas only 25 per cent of the eccentric systems survive merging . in figure
[ fig : collemax ] we show the fraction of binaries that experience a collision as a function of @xmath29 .
the two figures [ fig : cumulative ] and [ fig : collemax ] show similar trends ; high initial eccentricities enhance the collision rate .
.,width=264 ] in figure [ fig : branchemax ] we show the proportion of ionized systems and systems like @xmath23 which resulted from the standard run .
there seems to be a slight preference for initial eccentricities in the range @xmath6 . .
the solid line gives the results for encounters in which one binary was ionized .
the dashed line gives results for a subset of these , namely the ones which lead to systems like @xmath23 ( see @xmath17 [ sec : initial ] for the specifics of this encounter).,width=264 ] to further study the dependence of the collision probability on the stellar radii we define @xmath36 , the dimensionless effective radius for collisions . in figure [ fig : beta ] we show the branching ratios for collisions , ionizations and the sub - type of @xmath23 as a function of @xmath37 .
changing the stellar radii ( or @xmath37 ) by a factor of a few does not effect the collision rate much , and therefore the choice of the stellar radii is not critical to the results . .
our system of interest is represented with the dashed line .
, width=264 ]
until now we have considered the implications of a binary - binary encounter on the dynamical properties of @xmath20 , @xmath21 and @xmath23 .
the current observations of the binary allow us to constrain its evolution in the @xmath38myr elapsed from the moment of the encounter till now .
there are at least three further aspects that have to be integrated into a wholly plausible model of the interaction under consideration .
they are : 1 .
@xmath23 might be a triple system , and possibly even quadruple or quintuple ( see @xmath17 [ sec : speckle ] ) ; 2 .
stellar evolution models make it difficult to confirm that @xmath39 is of the same age as any of the other three stars ( see @xmath17 [ sec : age ] ) ; 3 .
tidal friction must operate within the present @xmath23 system , and we have to check whether the observed eccentricity can have been maintained for the 2.5@xmath25myr since the collision ( see @xmath17 [ sec : tidal ] )
. we deal with these in turn .
the spectroscopic binary @xmath23 is the brightest ( component a ) of three members of the multiple star ads 4193 .
the other two members ( b and c ) are @xmath40 and @xmath41 distant . at the distance of @xmath23 ( @xmath42pc , see table [ tab : vel ] )
these are quite widely separate , and for the time being we discount them as gravitationally bound companions . but ads 4193a is also a speckle binary , chara 250aa / ab , with a separation of @xmath43 ( mason et al .
this suggests an orbital period @xmath44yr . the brighter component ( aa ) of this speckle pair is the spectroscopic binary . strictly speaking
we should refer to the o9iii and b0.8iii - iv components of this @xmath45d binary as @xmath46 and @xmath47 , and we do so from now on .
the close speckle companion raises at least two questions .
is it reasonable that the companion was carried along during the four - body ( or rather 5-body ) encounter that we have supposed ? and could it affect the eccentricity of the spectroscopic orbit , by the mechanism of kozai cycles ( kozai 1962 ) ? we have not attempted to model 5-body encounters because of the enormous parameter space that they can be drawn from .
we can imagine it would be possible for the companion to remain bound after an encounter but we doubt the energy released would be enough to eject @xmath20 and @xmath21 with high speed .
it might be worth investigating this possibility in the future and testing these predictions . in a triple system
in which the outer orbit is fairly highly inclined to the inner orbit , the third body can have a major long - term effect on the inner pair , causing its eccentricity to cycle between a low and a high value .
for instance , if the mutual inclination is @xmath48 , @xmath49 can cycle between 0 and 0.76 , or between 0.5 and 0.86 ( kiseleva , eggleton & mikkola 1998 ) .
the period of the cycle is @xmath50yr in this case .
if the triple were formed early in the life of the cluster by some random encounter , an inclination of @xmath48would be quite likely ; alternatively , the kind of encounter that we think is needed to create the binary @xmath51 would probably randomize its inclination relative to the ( @xmath52 ) speckle pair .
thus it is possible that the observed high eccentricity of the spectroscopic binary @xmath51 is not a result of the encounter , but of the third body @xmath53 .
we now use stellar evolution model to constrain the ages of the four stars .
figure [ fig : hr ] is a theoretical hr diagram for masses in the range required .
the positions of all four stars , according to their effective temperatures and gravities as given by bagnuolo et al .
( 2001 ) , are indicated .
two isochrones are indicated , for 5 and 10 myr .
it is clear that the apparent ages of @xmath46 and @xmath47 are substantially different : we estimate 5 and 10 myr respectively .
the uncertainties appear to be of the order of @xmath54myr , but could be substantially greater on account of unquantifiable systematic errors .
the possibility that discrepant ages can be accounted for by roche - lobe overflow can almost certainly be discounted by the high eccentricity of the system .
the ages of @xmath20 and @xmath21 are comparable to that of @xmath46 , though @xmath20 might be a little younger and @xmath21 a little older .
= 1.1 , 1.125, ... ,1.5 . also shown ( thick lines ) are isochrones for ages 5 myr and 10 myr
. sloping nearly - straight dotted lines are for log @xmath55=2.8 , 3.0, ... ,3.8 , starting from the upper right corner and ending near the zams .
approximate positions for @xmath46 , @xmath47 , @xmath20 and @xmath21 , based on their temperatures and gravities , are indicated by aa1 , aa2 , ae and @xmath21 respectively . in the scheme the epoch of formation of the eldest binary is labelled as @xmath56 , the moment of the encounter as @xmath57myr and the present epoch as @xmath58myr.,width=302 ] the trapezium cluster , from which the four stars appear to have been ejected , is a well studied young cluster , the core of the orion nebula cluster , which contains about a dozen ob stars and many more fainter stars .
it is estimated that these stars are @xmath59myr old ( herbig & terndrup 1986 ; brown , de geus & de zeeuw 1994 ; prosser et al .
stars later than b2 are often though not always above the main sequence .
we may therefore wonder if the two giants of @xmath23 are still evolving _ towards _ the main sequence rather than away from it .
this seems improbable , because their ejection from the cluster @xmath5myr ago is amply long enough for contraction to the main sequence at their high luminosities .
this suggests , incidentally , that not _ all _ the trapezium stars are as young as @xmath60myr , and we appear to require that some massive stars have been forming over the last @xmath61myr .
an age spread from 1 to 10@xmath25myr has already been observed for stars @xmath62 ( palla & stahler 1999 ) , and we suggest that there may be a similar spread for ob stars .
an alternative possibility is that the parent cluster is one of the older subgroups in orion such as subgroup 1a or 1c ( see brown et al .
these subgroups have a kinematics very similar to the one of subgroup 1d ( which contains the trapezium cluster ) and overlap in position on the sky with @xmath23 .
the uncertainties in the theoretical hr diagram are of course substantial , and hard to quantify .
the models were computed using the code of pols et al .
( 1997 ) , which was found to give good agreement with the observational data of andersen ( 1991 ) .
the theoretical uncertainties have to be convolved with observational uncertainties but very substantial error must be involved if @xmath47 is to be of the same age as any of the other three stars .
a rather long shot is the possibility that the speckle companion is in fact the missing @xmath61myr - old companion of @xmath39 .
this would require a binary - triple interaction of the form : @xmath63 @xmath64 it is hard to assess the likelihood of this but in any event we should not ignore the possible existence of a fifth body in a complete analysis .
an important aspect to consider in the post - encounter evolution is the effect of tidal friction on the spectroscopic orbit .
the sum of the radii that we infer is @xmath65 of the periastron separation , which would seem sufficient for tidal effects to be important .
we have used the equilibrium - tide model of hut ( 1981 ) as further developed by eggleton , kiseleva & hut ( 1998 ) .
this model was successfully tested on the smc radio - pulsar binary 0045 - 7319 ( eggleton & kiseleva ( 2001 ) ) .
the orbital evolution expected since the encounter 2.5@xmath25myr ago is shown in figure [ fig : tidal ] .
we find that we could reach the present eccentricity and orbital period if we started with @xmath66=110 days and @xmath49=0.91 .
a high eccentricity is common for binaries which undergo exchange - ionization encounters as the final eccentricity distribution is expected to be thermal ( see @xmath17 [ sec : point ] ) . on the other hand , a long - period binary as a final outcome implies an even wider initial target binary , which means a lower binding energy available in the encounter .
it is unlikely to obtain recoil velocities as high as 100@xmath3 with a pre - encounter total energy which is smaller than the current one .
a tight binary ( with an orbital period similar to the observed one ) is needed in order to reproduce the velocities of the single runways .
figure [ fig : tidal ] shows that parallelization ( the alignment of the spin axis of the two binary components ) and pseudo - synchronization took @xmath67myr to achieve .
given the uncertainty in the viscous time - scale at least a factor of 10 it is possible that they have actually not yet been achieved .
the pseudo - synchronous rotation period ( hut 1981 ) is a fraction of the orbital period dependent only on eccentricity : the factor is 9.51 for @xmath49=0.76 , i.e. @xmath68d currently . with the radii computed from the evolutionary tracks of figure [ fig : hr ] , the rotation speeds we expect are @xmath69 and 115@xmath70 for the primary and the secondary respectively .
these values are somewhat greater than the values ( 110 and 70@xmath70 ) quoted by marchenko et al .
( 2000 ) , and may indicate that pseudo - synchronism has not yet been reached .
however we have already noted that the high eccentricity of the spectroscopic binary may not in fact have been generated by the collision but may be an artifact of kozai cycles driven by the third ( speckle ) body ( see @xmath17 [ sec : speckle ] ) . in kozai cycles , while the eccentricity fluctuates the semimajor axis is unperturbed ( to lowest order ) , and as a result the two components will in the long term seldom be as close together as they are at the periastra of the current orbit .
thus tidal friction might be negligible in this case .
the possibility that the eccentricity is due to the third body rather than to the interaction could mean that we have to revise our earlier conclusion that the apparently non - coeval nature of the spectroscopic binary is not due to roche lobe overflow , since the main argument against there having been any roche lobe overflow is that the orbit is eccentric
. however kozai cycles are normally suppressed if the quadrupolar distortion of the stars in the close binary is large ; and it usually _ is _ large enough as a star approaches roche lobe overflow .
once a star fills its roche lobe it normally grows larger still , until a very late stage of evolution when it has lost @xmath71 of its mass .
thus once kozai cycles are suppressed they are unlikely to re - establish themselves till this late stage , when the donor would be a small helium star rather than a b giant .
we hope that in the near future more information will be available on the speckle orbit , which should move appreciably in @xmath72yr
. the spectroscopic orbit may also be on the margin of detectability by speckle or adaptive optics ; a direct measure of its inclination could limit the possibilities substantially
. it will be desirable to search for the speckle companion in the spectrum of @xmath23 .
if it can be detected it may modify significantly the spectroscopic elements of the 29@xmath25d orbit .
we tried to reconstruct the event that ejected the runaways @xmath20 and @xmath21 and the binary @xmath23 from their parent cluster about 2.5@xmath25myr ago .
we here summarize different models that could explain the high velocities of the single stars and the discrepant ages of the binary components . *
the single stars and the binary are all unrelated . as shown by gies and bolton ( 1986 ) and hoogerwerf et al .
( 2000 , 2001 ) , this can be excluded on the basis of their kinematical and evolutionary properties .
the same dynamical encounter must have ejected the stars from the trapezium about 2.5 @xmath25myr ago . in order to explain the high velocities of the single stars and the age difference in the binary
, the encounter must have involved an ionization and an exchange . *
the stars were ejected as a result of a binary - binary interaction that occurred in the trapezium cluster about 2.5@xmath25myr ago .
this model is based on the observation that the velocity of the 4-body system is comparable to the velocity of the trapezium cluster .
the velocities of the stars and the period of the binary after the encounter are consistent with the observed values , so that we can conclude that this model can well reproduce the kinematics of the 4-body system .
if we consider stars with physical radii , a fraction of encounters result in a collision between two stars .
we thus expect to find collision products in dynamically active young clusters like the trapezium .
the fraction of collisions strongly depends on the initial eccentricity of the two binaries and therefore we consider it likely that @xmath23 , @xmath20 and @xmath21 were ejected in an encounter between two low - eccentricity binaries . in this case , the final properties of the system like the recoil velocities of the stars and the orbital period and eccentricity of @xmath23 do not differ from the ones obtained for highly eccentric initial binaries .
the occurrence rate of this type of encounter , once corrected for the collision probability , is about one in 10 myr .
+ a possible difficulty in this model is the short circularization time - scale for the orbit of the binary after the encounter .
our calculations show that only extremely eccentric post - encounter orbits ( @xmath73 evolve in an eccentricity like that of @xmath23 after 2.5 myr .
+ in addition , the model relies on the assumption that @xmath39 and @xmath21 are in the same binary before the encounter , implying that the two stars must be coeval .
our evolutionary calculations show that @xmath21 might be too hot , or equivalently too early in spectral type , to be on the same isochrone as @xmath39 . however , our knowledge about the temperatures and gravities of the stars is sufficiently imprecise that we can assume coevality for all four stars . *
the stars were ejected as a result of a triple - binary interaction .
this model is based on the possibility that @xmath23 is the brightest component of a speckle pair , making it a stable hierarchical triple .
this would require a 5-body encounter with the same total energy as our standard 4-body encounter resulting in the ejection of a triple and two single stars . according to this model
, the high eccentricity of @xmath23 may be a result of kozai cycles driven by the third body .
we havent tested this possibility but we can argue that this type of encounter would be rather rare and not energetic enough to explain the velocities of the runaways
. it may certainly be worth investigating this scenario to test these predictions as well as the ones about corotation and pseudo - synchronisation .
the last word has evidently not been said on the interesting interaction between @xmath23 , @xmath20 and @xmath21 , but we think it remains a fascinating combination of orbital dynamics , cluster dynamics , stellar evolution and tidal effects .
in addition to more elaborate numerical simulations , high accuracy parameters for the stars are needed in order to settle the co - evality issue .
new interferometric speckle observations , possibly resolving the spectroscopic binary too , will be crucial for a better understanding of the dynamical encounter that ejected the runaway stars .
we would like to thank douglas gies , william bagnuolo , lex kaper , anthony brown , ronnie hoogerwerf and michela mapelli for interesting discussions and comments on the manuscript .
this work was supported by the netherlands organization for scientific research ( nwo ) , the royal netherlands academy of arts and sciences ( knaw ) and the netherlands research school for astronomy ( nova ) .
this work was performed under the auspices of the u.s .
department of energy , national nuclear security administration by the university of california , lawrence livermore national laboratory under contract no . w-7405-eng-48 .
andersen , j. 1991 , a&a rev .
, 3 , 91 bacon , d. , sigurdsson , s. & davies , m. b. 1996 , mnras , 281 , 830 bagnuolo , w. g. , gies , d. r. & penny , l. 1994 , asp conference , 56 bagnuolo , w. g. , riddle , j. r. l. , gies , d. r. & parry , d. j. 2001 , apj , 554 , 362 blaauw , a. & morgan , w. w. 1954 , apj , 119 , 625 blaauw , a. 1961 , bull .
netherlands , 15 , 265 brown , a. g. a. , de geus , e. j. & de zeeuw , p. t. 1994 , a&a , 289 , 101 brown , a. g. a. , walter , f. m. & blaauw , a. 1998 , astroph/9802054 colin , j. p. , charles , r. e. , leonard , p. j. t. & frail , d. a. 1996 , aj , 111 , 1220 cordes , j. m. & chernoff , d. f. 1998 , apj , 505 , 315 duquennoy , a. & mayor , m. 1991 , a&a , 248 , 485 eggleton , p. p. & kiseleva , l. g. 2001 , apj , 562 , 1012 eggleton , p. p. , kiseleva , l. g. & hut , p. 1998 , apj , 499 , 853 gies , d. r. & bolton , c. t. 1986 , apjs , 61 , 419 gies , d. r. & lambert , d. l. 1992 , apj , 387 , 673 heggie , d. c. 1975 , mnras , 173 , 729 van den heuvel , e. p. j. , portegies zwart , s. , bhattacharya , d. & kaper , l. 2000 , a&a , 364 , 563 herbig , g. h. & terndrup , d. m. 1986 , apj , 307 , 609 herrero , a. , kudritzki , r. p. , vilchez , j. m. , kunze , d. , butler , k. & haser , s. 1992 , a&a , 261 , 209 hills , j. g. 1980 , apj , 240 , 242 hoogerwerf , r. , de bruijne , j. h. j. & de zeeuw , p. t. 2001 , a&a , 365 , 49 hoogerwerf , r. , de bruijne , j. h. j. & de zeeuw , p. t. 2000 , apjl , 544 , l133 hut , p. 1981
, a&a , 99 , 126 hut , p. & bahcall , j. n. 1983 , apj , 268 , 319 kaper , l. et al .
1997 , apj , 475 , l37 kiseleva , l. g. , eggleton , p. p. & mikkola , s. 1998 , mnras , 300 , 392 kopal , z. 1959 , close binary systems ( london : chapman and hall ) kozai , y. 1962 , aj , 67 , 591 leonard , p. j. t. & duncan , m. j. 1988 , aj , 96 , 222 leonard , p. j. t. 1990 , jrasc , 84 , 216 leonard , p. j. t. & dewey , r. j. 1993 , asp conference , 45 lyne , a. g. , lorimer , d. r. 1994 , nature , 369 , 127 makino , j. & aarseth , s. j. 1992 , pasj , 44 , 141 marchenko , s. v. et al .
2000 , mnras , 317 , 333 mason , b. d. et al .
1998 , aj , 115 , 821 mccaughrean , m. j. & stauffer , j. r. 1994 , aj , 108 , 1382 mcmillan , s. l. w. & hut , p. 1996 , apj , 467 , 348 meynet , g. , mermilliod , j. c. & maeder , a. 1993 , a&a , 98 , 477 palla , f. & stahler , s. w. 1999 , apj , 525 , 772 penny , l. r. 1996 , apj , 463 , 737 pols , o. r. , tout , c. a. , schrder , k .- p . , eggleton , p. p. & manners , j. 1997 , mnras , 289 , 869 portegies zwart , s. , verbunt , f. 1996 , a&a , 309 , 179 portegies zwart , s. 2000 , apj , 544 , 437 portegies zwart , s. , mcmillan , s. l. w. , hut , p. & makino , j. 2001 , mnras , 321 , 199 prosser , c. f. et al .
1994 , apj , 421 , 517 rasio , f. a. , mcmillan , s. & hut , p. 1995 , apj , 438l , 33 schaller , g. , schaerer , d. , meynet , g. & maeder , a. 1992 , a&as , 96 , 269 sigurdsson , s. & phinney , e. s. 1993 , apj , 415 , 631 stickland , d. j. , pike , c. d. , llyoyd , c. & howarth , i. d. 1987 , a&a , 184 , 185 stone , r. c. 1979 , apj , 232 , 520 stone , r. c. 1991 , aj , 101 , 333 tauris , t. m. & takens , r. j. 1998 , a&a , 330 , 1047 turon , c. et al .
1992 , the hipparcos input catalogue , esa , sp-1136 zwicky , f. 1957 , morphological astronomy ( berlin : springer )
the procedure described in section @xmath17[sec : code ] for the computation of the maximum impact parameter in each scattering experiment is incorporated in the actual computation of cross - sections .
the cross - section for an event x is derived from ( mcmillan & hut 1996 ) @xmath74 where @xmath75 is the total number of scatterings resulting in an outcome x in the shell @xmath76 and @xmath77 is the total number of scatterings performed in the shell @xmath76 .
the uncertainty in the cross - section is given by @xmath78 in analogy to the case of binary - single star scattering ( hut & bahcall 1983 ; sigurdsson & phinney 1993 ) and in order to compute physical time - scales for different type of interactions , we define a cross - section @xmath79 normalized to the geometric cross - section and corrected for gravitational focusing .
a characteristic time - scale for a process x , defined as the mean time between subsequent occurrences of x , can be evaluated from ( bacon , sigurdsson & davies 1996 ) @xmath80 where @xmath81 represents the mean stellar density in the cluster and @xmath82 is the fraction of binaries in the core .
during the numerical integration of a binary - binary encounter the status of the system is analysed after each 20 orbital periods of the initial target binary and the simulation is stopped if one of the following criteria is met : * only two stars remain , in which case their orbits are determined and the system is classified as a binary if it is bound or as two single stars if it is unbound ; * three stars remain , in which case the system is considered ionized if the total energy is positive , the stars are widely separated and receding from each other while it is considered as a binary and a single star if the total energy is negative but the single star is escaping from the binary center of mass ; * the 4-body system is split in binaries whose centers of mass are unbound , widely separated and receding . in the other cases
the system is considered bound and the integration continues until a stopping criterion is satisfied or a maximum integration time is reached .
|
we study the dynamical interaction in which the two single runaway stars @xmath0 and @xmath1 and the binary @xmath2 acquired their unusually high space velocity .
the two single runaways move in almost opposite directions with a velocity greater than 100@xmath3 away from the trapezium cluster .
the star @xmath2 is an eccentric ( @xmath4 0.8 ) binary moving with a velocity of about 10@xmath3 at almost right angles with respect to the two single stars .
the kinematic properties of the system suggest that a strong dynamical encounter occurred in the trapezium cluster about @xmath5myr ago .
curiously enough , the two binary components have similar spectral type but very different masses , indicating that their ages must be quite different .
this observation leads to the hypothesis that an exchange interaction occurred in which an older star was swapped into the original @xmath2 binary .
we test this hypothesis by a combination of numerical and theoretical techniques , using n - body simulations to constrain the dynamical encounter , binary evolution calculations to constrain the high orbital eccentricity of @xmath2 and stellar evolution calculations to constrain the age discrepancy of the two binary components .
we find that an encounter between two low eccentricity ( @xmath6 ) binaries with comparable binding energy , leading to an exchange and the ionization of the wider binary , provides a reasonable solution to this problem .
[ firstpage ] methods : n - body simulations - stars : binaries : spectroscopic - stars : individual : hd 34078 , hd 37043 , hd 38666
|
the achievement of bose - einstein condensation in 1995 in dilute gases of @xmath0rb @xcite , @xmath1li @xcite and @xmath2na @xcite revolutionised atomic physics . since that time , bose - einstein condensation has been achieved for several other alkali metal species ( @xmath3rb @xcite , @xmath4k @xcite and @xmath5cs @xcite ) and a few other systems ( @xmath6h @xcite , metastable he @xcite , @xmath7yb @xcite and @xmath8cr @xcite ) .
intense effort has been devoted to the study of the new properties of bose - einstein condensates @xcite .
the field was further broadened by the achievement of quantum degeneracy in fermi gases of @xmath9k @xcite and @xmath10li @xcite , and ultracold fermionic quantum matter has proved to exhibit a new range of novel properties .
bose - einstein condensation and fermi degeneracy in dilute gases typically require temperatures between 1 nk and 1 @xmath11k .
however , new quantum properties start to appear at temperatures around 1 mk , where de broglie wavelengths become large compared to atomic and molecular dimensions . under these circumstances
, collisions are fully quantum - mechanical and are primarily sensitive to long - range interactions .
the region below 1 mk is generally referred to as the _ ultracold _ regime . over the last few years
, the focus of research in quantum matter has shifted to the _ control _ of ultracold quantum systems .
a particularly important development has been the ability to form and manipulate _
molecules _ in ultracold atomic gases .
molecules have a much richer energy level structure than atoms , and offer many new possibilities for quantum control .
perhaps most importantly , _
dipolar _ molecules interact with one another much more strongly and at longer range than atoms .
dipolar quantum gases are predicted to exhibit more new features @xcite and have possible applications in quantum computing @xcite .
cold molecules also have applications in high - precision measurement , and high - resolution spectroscopy on cold molecules may allow the measurement of fundamental physical properties such as the electric dipole moment of the electron @xcite , the energy differences between enantiomers ( which result from parity violation ) @xcite and the time - dependence of the fine - structure constant @xcite .
there are two approaches to producing molecular quantum gases : _ direct _ approaches , in which pre - existing molecules are cooled from room temperature , and _ indirect _ approaches , in which molecules are formed from precooled atoms .
the direct approaches have been reviewed previously @xcite and there are also several reviews that focus on the applications of scattering theory to directly cooled molecules @xcite . the purpose of this article is to review molecule formation in ultracold quantum gases by _ indirect _ methods .
the physics of cooling and trapping atoms @xcite and of the production and properties of bose - einstein condensates @xcite has been reviewed many times .
we will restrict ourselves here to a brief discussion , focussing on aspects of the subject that are essential to understanding the present review but may be unfamiliar to readers with a background in physical chemistry rather than atomic physics .
all fundamental particles are either bosons or fermions .
bosons have integer spin quantum numbers , while fermions have odd half - integer spin quantum numbers .
the fundamental difference is encompassed by the _ pauli principle _ , which states that the total wavefunction for a system must be symmetric with respect to exchange of any pair of identical bosons , but antisymmetric with respect to exchange of a pair of identical fermions .
the most important consequence of the pauli principle is the _ pauli exclusion principle _ , which states that no two fermions in the same spin state can occupy the same spatial state . under circumstances where
individual electrons can not be exchanged , atoms and molecules are _
composite _ bosons or fermions . any atom or molecule with an even number of nucleons and electrons
is a composite boson and any with an odd number is a composite fermion . for the alkali metals , with an odd number of electrons , isotopes with bosonic nuclei ( @xmath10li , @xmath9k ) are composite _ fermions _ and isotopes with fermionic nuclei ( @xmath1li , @xmath2na , @xmath4k , etc . )
are composite _
bosons_. an alkali metal atom with nuclear spin @xmath12 in its ground electronic state ( @xmath13s@xmath14 ) can have total angular momentum @xmath15 . in a magnetic field ,
the energy levels split into components labelled by the projection quantum number @xmath16 , as shown for @xmath3rb in fig .
[ rb85-zeeman ] ( the breit - rabi diagram ) .
an atom for which the ground state has the lowest value of @xmath17 is said to have _ regular _ hyperfine structure , and one in which the order is reversed is said to have _ inverted _ hyperfine structure .
the only alkali metal atom important to the present review that has inverted hyperfine structure is @xmath9k .
the projection quantum number @xmath16 is a good quantum number at any field , but @xmath17 is conserved only at zero field . at fields above the avoided crossings in fig .
[ rb85-zeeman ] , @xmath17 no longer describes the character of the states . in this region
the nearly conserved quantities are the individual projections @xmath18 and @xmath19 .
this occurs at quite high field for @xmath3rb but at much smaller fields for atoms with smaller hyperfine splittings such as @xmath10li .
nevertheless , @xmath20 always provides a unique _ label _ for a hyperfine state by following the curve back to low field .
[ htbp ] rb in a magnetic field .
@xmath3rb has @xmath21 and @xmath22 and 3 .
figure from tiesinga @xcite.,width=321 ] molecules may have more than one nucleus with non - zero spin , and have mechanical as well as electronic angular momentum .
their energy level structure is thus considerably more complex than for atoms .
nevertheless , it remains true that the zero - field levels are characterised by a total angular momentum @xmath17 , and that in a field these are split into components with projection quantum number @xmath16 .
ultracold atoms and molecules would condense if they came into contact with the walls of a vessel .
it is therefore necessary to trap them without physical walls . in most experiments ,
an atomic beam is first decelerated using light pressure in a zeeman slower @xcite , which maintains the atoms in resonance as they slow down .
the slow atoms are then confined in a magnetic @xcite or magneto - optical @xcite trap .
levels whose energy _ increases _ in a magnetic field ( see fig.[rb85-zeeman ] ) are known as _ low - field - seeking _ , and those whose energy _ decreases _ in a field are known as _
high - field - seeking_. a magnetic trap @xcite operates by creating a local minimum in the magnetic field strength @xmath23 , so that atoms in low - field - seeking states are trapped . since it is not possible to create a local _ maximum _ in @xmath23 in free space , high - field - seeking states can not be trapped magnetically . for alkali metal atoms ,
a magnetic trap has a typical depth of around 1 mk . a magneto - optical trap ( mot ) @xcite uses a combination of magnetic fields and doppler cooling @xcite to trap atoms at considerably higher densities than is possible with magnetic fields alone .
a mot can also cool atoms below the theoretical limit of doppler cooling by a mechanism known as sisyphus cooling @xcite . for example , in early work on @xmath2na , lett _ et al . _
@xcite observed a temperature of 40 @xmath11k , compared to the 240 @xmath11k expected .
even this is not by itself sufficient to achieve bose - einstein condensation , and a final stage of cooling , often by evaporative cooling @xcite , is needed to achieve sub-@xmath11k temperatures .
ultracold atoms can also be confined in an optical dipole trap @xcite .
optical dipole traps have the advantage that all magnetic sublevels can be trapped simultaneously .
they rely on the fact that the energy of an atom in an electric field oscillating at frequency @xmath24 is @xmath25 , where @xmath26 is the electric field strength .
the frequency - dependent polarizability @xmath27 is positive at low frequencies but is enhanced ( and changes sign ) near absorption frequencies .
an optical dipole trap operates by creating an electric field _
maximum _ in a region of strong laser radiation , and can be either _ near - resonant _ , taking advantage of the enhancement in @xmath27 near an absorption , or _
far off - resonance_. a far off - resonance trap ( fort ) @xcite causes less heating but is much shallower than a near - resonant trap ( sometimes as little as 10 @xmath11k ) .
an optical dipole trap in which the laser frequency is so low that the polarizability is close to its static value is referred to as a quasi - electrostatic trap ( quest ) @xcite .
an optical dipole trap can confine molecules as well as atoms @xcite .
in addition , molecules with magnetic dipole moments can be trapped magnetically @xcite .
molecules with electric dipole moments can be trapped by an analogous _ electrostatic _ approach @xcite .
most traps produce a trapping potential that is nearly harmonic near the minimum . because of this ,
the `` translational '' energy spectrum of trapped atoms is not actually continuous , as is the case for free particles .
instead , it is discretised by the confinement in the trap : the energy level spacings @xmath28 are typically 10 to 1000 hz , corresponding to 0.5 to 50 nk .
bosons and fermions follow quite different quantum statistics : bose - einstein and fermi - dirac statistics respectively . in a bose - einstein condensate ,
nearly all the atoms are in the lowest level in the trap ( though there are always uncondensed atoms coexisting with the condensate ) . in a harmonic trap
, condensation occurs at a critical temperature @xmath29 given approximately by @xmath30 , where @xmath31 is the number of atoms @xcite . in a fermi - degenerate gas , by contrast , each level can accommodate only one atom in each spin state ( @xmath32 ) , so that the system is characterised by full occupation of trap levels up to the fermi level , forming a `` fermi sea '' .
the fermi temperature @xmath33 is again proportional to the level spacing , @xmath34 @xcite .
both bose - einstein condensation and fermi degeneracy typically occur at temperatures below 1 @xmath11k .
the scattering wavefunction for collision of a pair of structureless atoms is conveniently written @xmath35 .
the radial wavefunction @xmath36 obeys the 1-dimensional schrdinger equation , @xmath37\chi(r)=0,\ ] ] where @xmath38 is the total energy , @xmath39 is the effective potential energy and @xmath11 is the reduced mass . for two atoms colliding with zero kinetic energy , @xmath40 as @xmath41 , so @xmath36 has zero curvature in that region and becomes linear in the interatomic distance @xmath42 , @xmath43 the atom - atom interaction is characterised at the simplest level by the _ scattering length _ , @xmath44 , which is the distance at which the continuation of the asymptotic straight line ( [ asymptwavefun ] ) crosses zero .
many of the properties of a bose - einstein condensate depend only on the scattering length and are unaffected by short - range properties of the wavefunction such as the number of nodes .
the chemical potential @xmath45 of a uniform bose gas is proportional to @xmath44 , @xmath46 , where @xmath47 is the number density , @xmath48 and @xmath49 is the atomic mass @xcite .
a positive scattering length thus corresponds to an interaction that is overall repulsive , while a negative scattering length corresponds to an interaction that is overall attractive .
a large bose - einstein condensate can usually exist only for positive scattering lengths ; for negative scattering lengths there is a limit to the size of condensate that can be formed @xcite .
the scattering length is closely related to the energy of the highest bound state of the atom - atom pair , @xmath50 .
if @xmath50 is small and negative ( a bound state just below threshold ) , the scattering length is large and positive and is related to the bound - state energy approximately by @xmath51 the scattering length increases to infinity as @xmath50 approaches zero and reappears at large negative values when @xmath52 .
the scattering length depends on the interaction potential and on the reduced mass .
it is thus different for pairs of atoms interacting on singlet and triplet curves , and indeed for different hyperfine states .
it is also different for different isotopic species .
a very important discovery was that the interactions between ultracold atoms can be tuned using magnetic fields @xcite .
as described above , an atom with nuclear spin @xmath12 in a @xmath13s@xmath14 state can have total angular momentum @xmath15 .
when two such atoms interact , there are 3 closely - spaced thresholds corresponding to different combinations of hyperfine states as shown in fig .
[ feshbach - curves ] , and there are sets of vibrational levels correlating with each threshold .
some of the high - lying vibrational levels of the upper curves can lie above the lower thresholds .
vibrational levels embedded in a continuum are quasibound and produce feshbach resonances @xcite . in zero field , such resonances are characterised by their energy @xmath53 and width @xmath54 ( in energy space ) .
[ htbp ] rb@xmath55 ( with nuclear spin @xmath21 ) showing the 3 hyperfine thresholds
. also shown is a zero - energy scattering wavefunction on the lowest curve @xmath56 and a bound - state wavefunction on the @xmath57 curve that is above the lowest threshold and produces a feshbach resonance . reprinted with permission from courteille _
et al . _
copyright 1998 by the american physical society.,width=321 ] each combination of atomic quantum numbers produces a _ channel _ , and in general there are several channels correlating with each threshold ( corresponding to different values of projection quantum numbers @xmath16 and values of the quantum numbers @xmath58 and @xmath59 that describe mechanical rotation of the atoms about one another ) .
each channel has its own potential energy curve , though for simplicity only one curve is shown for each threshold in fig .
[ feshbach - curves ] . at energy @xmath38 , each channel is described as either _ open _ or _ closed _ ( energetically accessible or inaccessible at @xmath60 ) .
for example , at the energy of the state shown in fig .
[ feshbach - curves ] , channels corresponding to the lowest threshold are open and the rest are closed .
[ thbp ] rb@xmath55 as a function of magnetic field .
feshbach resonances occur at the points marked with filled circles , where a molecular state crosses a threshold .
@xmath0rb has @xmath61 , so there are higher thresholds corresponding to @xmath62 and ( 2,2 ) .
reprinted with permission from marte _ et al . _
copyright 2002 by the american physical society.,width=321 ] in a magnetic field , both the atomic levels ( thresholds ) and the molecular levels split and shift as shown for @xmath0rb in fig.[rb87-tuning ] . a high - lying vibrational level correlating with one hyperfine state
can often be tuned across a lower threshold with an applied magnetic field . in work on ultracold gases , where the collision energy is typically fixed at very near zero and a magnetic field
can be varied , the term feshbach resonance " has come to be applied to the behaviour of scattering properties as a function of applied field as a state crosses threshold .
[ eqetopa ] applies even in the multichannel case , _ not _ just when the scattering is governed by a single potential curve .
the scattering length thus has a pole whenever there is a bound state at zero kinetic energy , as shown for @xmath5cs@xmath55 in the lower panel of fig .
[ cs - tuning - a+relax ] . as a function of magnetic field @xmath23 , the scattering length in the vicinity of a feshbach resonance has the form @xcite @xmath63 where @xmath64 is the background scattering length , @xmath65 is the resonance position ( defined as the field at which @xmath44 is infinite , which is not quite the same as the field at which the bound state is at zero kinetic energy ) and @xmath66 is related to the width of the resonance @xmath54 . in practice , it is of course possible for resonances to overlap or for a sharp resonance to occur in the wings of a broad one .
for example , the strong field - dependence of the scattering length for @xmath5cs@xmath55 at low magnetic field shown in fig .
[ cs - tuning - a+relax ] can be interpreted in terms of a broad feshbach resonance at @xmath67 g @xcite .
[ htbp ] cs collisions with @xmath68 as a function of magnetic field .
lower panel : scattering length based on theoretical parameters in ref .. upper panel : loss rate for radiative collisions @xcite .
note that some of the sharpest resonances are too narrow to see in the scattering length but are still observed in the loss rates .
figure from chin @xcite.,width=321 ] the elastic cross section @xmath69 for identical bosons at kinetic energy @xmath70 is approximately given by @xcite @xmath71 so that @xmath69 passes through a peak of height @xmath72 ( corresponding to @xmath73 ) at a resonance . other collisional properties also show sharp features .
for example , for @xmath5cs the radiative loss rate ( which is due to atom pairs temporarily excited to the excited electronic state ) exhibits sharp peaks as shown in the upper panel of fig .
[ cs - tuning - a+relax ] @xcite .
magnetic tuning of feshbach resonances was first observed by inouye _
et al . _
@xcite and stenger _
et al._@xcite , who detected enhanced loss rates from a bose - einstein condensate of @xmath2na as a function of magnetic field , and by courteille _
et al . _
@xcite , who observed enhanced photoassociation rates in the vicinity of a feshbach resonance in @xmath3rb .
the loss rates were attributed either to excitation of atoms to higher trap states during the field ramp @xcite or to inelastic collisions involving the transiently formed molecular state @xcite . in experiments on @xmath3rb , roberts _ et al . _
@xcite and donley _
et al._@xcite demonstrated that tuning the scattering length suddenly from positive to negative could cause controlled collapse of a bose - einstein condensate .
the collapse produces an explosion of atoms from the condensate and has come to be known as a bosenova by analogy with astrophysical supernovae .
the possibility of using feshbach resonances to create ultracold molecules was first predicted in 1999 @xcite . as described above
, a feshbach resonance occurs when a molecular level crosses an atomic threshold as a function of the magnetic field @xmath23 . in reality , since the atomic and molecular states are coupled , there is an avoided crossing .
thus if the magnetic field is tuned across the resonance , slowly enough to follow the avoided crossing adiabatically , pairs of trapped _
atoms _ can be converted into trapped _
molecules _ as shown in fig .
[ avoided ] .
since no kinetic energy is created in the process , the molecules are created with essentially the same energy as the atoms .
[ htbp ] the actual arrangement of states in a feshbach resonance is sometimes the mirror image of that shown in fig .
[ avoided ] . in the following discussion , the side of the resonance where molecules are lower in energy than atoms will be referred to as the _ molecular side _ of the resonance .
the first signatures of trapped molecules produced by feshbach resonance tuning were observed by donley _
et al._@xcite in 2002 .
they worked with magnetically trapped @xmath3rb , so were restricted to low - field - seeking states .
@xmath3rb has nuclear spin @xmath21 , and donley _
et al._worked with trapped atoms in the @xmath74 state , which is not the lowest state in a magnetic field ( see fig.[rb85-zeeman ] ) . in this case
the energy levels are the mirror image of those shown in fig .
[ avoided ] , with the molecular state below the atomic state on the _ high - field _ side of the resonance . the experiment of donley _ et al . _
was a little more complicated than a simple ramp over the resonance as shown in fig .
[ avoided ] .
since the background scattering length for this system is negative , @xmath75 is negative on the atomic side even far from the resonance .
for this reason it was not possible to approach the resonance from the atomic side , because the condensate collapsed @xcite .
et al . _
therefore used a magnetic field profile that approached the feshbach resonance from the _ molecular _ side but did not actually cross it .
the fast field ramp mixed the atomic and molecular states , and donley _
et al . _ observed quantum beats ( ramsey fringes ) between trapped atoms and molecules . over the following two years
, the technique was developed by several groups and extended to other bosonic systems : @xmath5cs@xmath55 @xcite , @xmath0rb@xmath55 @xcite and @xmath2na@xmath55 @xcite .
in all these experiments atoms were prepared in the _ lowest _ hyperfine state in a magnetic field , with @xmath76 and @xmath77 .
such states can not be trapped magnetically ( because they are high - field - seeking ) , so the atoms were confined in optical traps .
molecules were created by sweeping the magnetic field _ across _ a resonance as shown in fig . [ avoided ] .
the molecules were separated from the remaining atoms either magnetically @xcite or by using a laser resonant with the atoms but not the molecules to push the atoms out of the trap @xcite .
the molecules were then detected and imaged by converting them back to atoms with a reverse field sweep .
fig.[cs2bec ] shows images of the cs@xmath55 molecular cloud created by herbig _
et al . _
@xcite .
[ htbp ] cs@xmath55 molecules .
a : magnetically levitated atomic bec ; b : levitated atomic bec with falling molecular cloud below ; c : levitated molecular cloud with rising atomic bec above . reprinted with permission from herbig
et al . _
copyright 2003 by macmillan publishers ltd.,width=321 ] in all these experiments involving bosonic atoms it was found that the molecules were lost from the trap within a few milliseconds .
the fast trap loss was attributed to atom - diatom collisions .
the molecules are formed in a very highly excited state , often the highest vibrational state that exists in the potential well .
there are always many lower - lying states , and even near dissociation the vibrational spacing is large compared to the depth of the trap .
thus inelastic atom - molecule collisions , @xmath78 or molecule - molecule collisions , @xmath79 always release enough kinetic energy to eject both collision partners from the trap .
it should be noted that the molecules are not _ destroyed _ in such collisions , but they are lost from the trap and are no longer ultracold . as remarked above , the molecules produced by donley _
et al._@xcite are formed from @xmath3rb atoms in the @xmath74 state , which is not the lowest state in a magnetic field .
they are thus not strictly bound , and can dissociate spontaneously without collisions to form atoms in lower - lying hyperfine states .
this has been studied experimentally by thompson _
et al . _
@xcite and theoretically by khler _
et al . _
et al . _
@xcite adapted the experiment of donley _
et al . _
@xcite to create @xmath3rb@xmath55 molecules using a magnetic field sweep through the feshbach resonance without holding the atomic condensate at @xmath80 for long enough for it to collapse . as in the earlier experiments ,
the molecules were formed in states above the lowest hyperfine threshold . even at densities where collisional loss was very slow , thompson _
et al . _ found that the molecules decayed within 1 ms at magnetic fields far from resonance .
however , close to resonance , they were able to achieve lifetimes of tens of milliseconds .
it should be noted that this decay mechanism is not applicable to the experiments on @xmath5cs@xmath55 , @xmath0rb@xmath55 and @xmath2na@xmath55 @xcite , where the molecules are formed in states that lie below the lowest atomic threshold .
molecules formed by feshbach resonance tuning are very large .
even the closed - channel part of the wavefunction corresponds to a molecule in a very high vibrational state , as shown in fig.[feshbach - curves ] ; the wavefunction peaks near the outer turning point , which can be at distances approaching @xmath81 .
however , even this underestimates the size of feshbach molecules .
et al . _
@xcite have investigated a model of the feshbach resonance used to produce @xmath3rb@xmath55 by donley _
et al . _
@xcite , and have shown that for magnetic fields near the resonance the molecular wavefunction is dominated by the _ open _ ( atomic ) channel and has a wavefunction that dies off as @xmath82 at long range ; the mean internuclear distance is on the order of @xmath83 , which near a resonance can be several thousand @xmath84 .
an example of this is shown in fig .
[ longrange ] .
[ htbp ] rb@xmath55 close to a feshbach resonance ( @xmath85 mt , @xmath86 ) and slightly further away ( @xmath87 mt , @xmath88 ) . reprinted with permission from khler _ et al . _
copyright 2003 by the american physical society.,width=321 ] in parallel to the work on boson dimers , molecules were created from pairs of _ fermionic _
atoms by tuning through feshbach resonances .
collision rates for two identical fermions are suppressed because s - wave scattering ( partial wave @xmath89 ) is forbidden .
accordingly , the fermion work focussed on pairs of atoms in different spin states , again in optical traps .
molecule formation was first achieved for @xmath9k@xmath55 by regal _
et al . _
@xmath9k has nuclear spin @xmath90 and an inverted hyperfine structure , so that the atomic ground state in a magnetic field has @xmath91 . regal _
et al._used a mixture of @xmath92 and @xmath93 atoms and formed molecules by magnetic tuning through a feshbach resonance .
once again the molecules were found to be short - lived ( @xmath94 ms ) .
the lifetime was attributed to vibrationally inelastic collisions @xcite , though spontaneous dissociation by spin relaxation is also possible in this case .
a major experimental breakthrough came in mid-2003 , when four groups @xcite independently reported within a period of 6 weeks that fermion dimers can be remarkably stable to collisions when the atom - atom scattering length is tuned to a large positive value .
@xmath10li has nuclear spin @xmath95 , and strecker _
et al._@xcite and cubizolles _
et al._@xcite prepared @xmath10li@xmath55 molecules by feshbach resonance tuning in mixtures of the lowest two spin states , correlating with @xmath96 and @xmath97 at low field .
both groups showed that the molecules remained trapped for 1 s or more before being dissociated by a reverse magnetic field sweep .
et al . _
@xcite prepared @xmath10li@xmath55 molecules by a different method , taking advantage of the dramatically enhanced 3-body recombination rate near a feshbach resonance , and observed comparable lifetimes .
et al . _
@xcite and jochim _
et al._@xcite showed that the lifetime was particularly large close to the resonance , where the scattering length is large and positive .
et al._@xcite carried out analogous experiments on @xmath9k@xmath55 , both for the spin states involved in their earlier experiments @xcite and for molecules formed from @xmath91 and @xmath98 .
they confirmed the fast decay far from resonance , but showed that for large positive scattering lengths the lifetime was dramatically enhanced . by the end of 2003 ,
three different groups had succeeded in creating long - lived molecular bose - einstein condensates of fermion dimers .
et al . _
@xcite and zwierlein _ et al . _
@xcite achieved this by evaporative cooling combined with 3-body recombination in a mixed gas of @xmath10li in its @xmath96 and @xmath97 states , held at large positive scattering length close to a feshbach resonance . under these circumstances
3-body recombination to form molecules is thermodynamically favourable and the molecules are long - lived .
et al . _
@xcite formed @xmath9k@xmath55 from a very cold fermi gas of @xmath9k@xmath55 atoms in their @xmath91 and @xmath98 states , using a feshbach ramp so slow that thermal equilibrium was maintained during the field sweep .
all three groups observed the sudden appearance of a sharp spatial peak in the density with decreasing temperature , as shown in fig .
[ k2bec ] for @xmath9k@xmath55 .
this is widely regarded as the smoking gun " of bose - einstein condensation @xcite .
[ htbp ] k@xmath55 after 20 ms of free expansion , above and below the critical temperature for bose - einstein condensation .
the condensed cloud ( right ) shows the tight spatial peak characteristic of a condensate .
reprinted with permission from greiner _
et al . _
copyright 2003 by [ check organisation],width=321 ] petrov _ et al . _
@xcite analysed the stability of fermion dimers in terms of the long - range form of the wavefunction . in the case where the atom - atom scattering length @xmath44 is much larger than the range of the atom - atom potential @xmath99
, they showed that both the atom - molecule and molecule - molecule inelastic collision rates are suppressed by fermi statistics .
however , their derivation applies only to molecules that are in long - range states , with a wavefunction that depends on the scattering length , @xmath100 .
as will be discussed in more detail below , cvita _ et al . _
@xcite have shown computationally that there is no systematic suppression of the atom - molecule inelastic rate for fermion dimers in low - lying vibrational levels , even when @xmath44 is large and positive .
a major reason for interest in fermion dimers is that they provide tunable models for studying problems in condensed matter physics such as the origin of superfluidity and superconductivity .
according to the widely accepted theory of bardeen , cooper and schrieffer ( bcs ) @xcite , superconductivity occurs because pairs of electrons are composite bosons and can therefore condense .
however , the pairs of electrons involved in superconductivity ( cooper pairs ) are much larger than the mean separation between electrons .
this is different from the usual regime of molecular bose - einstein condensation ( bec ) , because far from a resonance a fermion dimer is relatively strongly bound and is small compared to the typical separation between unbound atoms .
however , close to resonance the size " of the molecules increases and can become comparable to the atom - atom spacing . under these circumstances
the molecules " are interpenetrating and lose their identity in very much the same way as cooper pairs of electrons .
the beauty of the fermionic atom systems is that this transition can be followed as a function of magnetic field .
the bec - bcs crossover between a molecular bose - einstein condensate and a condensate of atomic cooper pairs has been studied extensively .
the bcs regime is an intrinsically many - body regime in which 2-body theories can not be expected to work . at the centre of a resonance
the 2-body scattering length is infinite but the pair size in a many - body system is limited to the atom - atom spacing . beyond the resonance , where the scattering length is negative , the binding forces are much smaller and
the pair size becomes much larger then the interparticle spacing .
et al._@xcite showed that the transition between the two regimes can be achieved smoothly and reversibly .
et al . _
@xcite and zwierlein _ et al . _
@xcite demonstrated bose - einstein condensation of cooper pairs in the bcs regime and measured the atomic momentum distribution in the bcs region . chin _
et al . _
@xcite and greiner _
et al._@xcite observed a pairing gap characteristic of superfluidity .
et al . _
@xcite have measured the closed - channel component of the pair wavefunction on both sides of a resonance and shown that , though small , it persists on the bcs side of the resonance , contrary to the predictions of 2-body theory .
other studies have focussed on signatures of superfluidity rather than cooper pairing .
et al . _
@xcite and bartenstein _ et al . _
@xcite observed collective oscillations in a strongly interacting fermi gas that suggest superfluid behaviour in the bcs regime , while kinast _ et al . _
@xcite measured the heat capacity of such a gas .
most recently , zwierlein _
et al._@xcite provided conclusive evidence of superfluidity by observing arrays of vortices on both the bec and bcs sides of a resonance in @xmath10li@xmath55 .
it is possible to trap two different alkali metal species simultaneously , and magnetic feshbach resonances have been observed in rbk @xcite and lina @xcite .
there is little doubt that tuning through such resonances will soon be used to form heteronuclear molecules .
however , it should be noted that neutral heteronuclear molecules in long - range states do _ not _ have significant dipole moments : typical values are less than 0.3 d at @xmath101 @xcite and decay as @xmath102 at long range @xcite .
it is in principle possible to form molecules larger than diatomic , either by direct association from atoms or by association of smaller molecules .
et al._@xcite have formed cs@xmath55 molecules by feshbach resonance tuning and then separated out the remaining atoms magnetically .
they observed field - dependent resonances in the inelastic loss rates that they attributed to cs@xmath103 bound states near the molecular scattering threshold .
[ htbp ] ) and for atom + diatom ( @xmath104 ) .
the characteristic factor of 22.7 in @xmath44 has been reduced to 2 for the purpose of illustration .
reproduced from kraemer _
et al._@xcite.,width=321 ] for three atoms there is the intriguing prospect of forming efimov states @xcite , which are long - range trimer states that exist even when the corresponding dimers are unbound .
indeed , efimov showed that , if the pair potential has exactly one bound state at zero energy ( corresponding to an infinite scattering length ) , there are an infinite number of such trimer states .
the helium trimer is predicted to have one efimov state @xcite , but it has not yet proved possible to observe this experimentally .
however , alkali metal atoms with tunable interactions offer new possibilities for a slightly different type of efimov state , which occurs whenever the scattering length is large but is complicated by the existence of a large number of deeply bound states .
braaten and hammer @xcite and nielsen _ et al . _
@xcite have shown that efimov states will cause resonant enhancements in 3-body recombination rates at characteristic values of the scattering length that differ by successive factors of 22.7 .
the characteristic dependence of energies on scattering length is shown in fig .
[ efimov ] .
very recently , kraemer _
et al._@xcite have measured trap loss in an ultracold gas of cs atoms as a function of scattering length , and observed a peak that they attribute to resonance between an efimov state of the trimer and the threshold for 3 separated atoms .
dimers can also be formed in cold atomic gases by photoassociation , as predicted by thorsheim _
et al . _ in 1987 @xcite .
developments up to 1999 were reviewed by stwalley and wang @xcite and more recent work by jones _
et al . _
the early work focussed on 1-photon photoassociation spectroscopy , forming molecules in electronically excited states . more recently ,
however , it has become possible to form ultracold molecules in their electronic ground states by 2-photon processes as shown in fig.[photoass-2colour ] .
this was first achieved by fioretti _
et al . _
@xcite , who photoassociated cs@xmath55 to an excited @xmath105 electronic state with a double minimum @xcite and observed ultracold molecules formed by spontaneous emission to the lowest triplet state .
et al . _
@xcite carried out a similar experiment to form the @xmath106 ground state of k@xmath55 and used resonant two - colour ionization to show that the vibrational distribution peaked at @xmath107 , just over half way up the ground state potential well .
they subsequently @xcite developed a 2-photon excitation scheme via an excited @xmath108 state that produced molecules even further down the ground - state well .
optical trapping of ground - state cs@xmath55 molecules was achieved by takekoshi _
et al . _
@xcite and magnetic trapping by vanhaecke _
_ @xcite .
[ htbp ] . reprinted with permission from boesten _
et al . _
copyright 1999 by institute of physics.,width=321 ] an additional degree of control can be introduced by using stimulated raman adiabatic passage ( stirap ) @xcite , in which a second laser detuned from the excitation laser brings the molecules down to bound levels of the ground electronic state @xcite .
this was first achieved by wynar _
et al . _
@xcite , who worked in a bose - einstein condensate of @xmath0rb and produced ultracold @xmath0rb@xmath55 molecules in a specific vibration - rotation ( and hyperfine ) state by stirap via the @xmath105 state . in their experiment
the dump laser was detuned by only 636 mhz from the pump laser , and the molecules were formed in the second - to - last vibrationally excited state .
producing molecules in this way has the major advantage that coherence can be maintained @xcite ; this is not the case if spontaneous emission is involved .
et al._@xcite have used 2-colour photoassociation to produce a coherent superposition of atomic @xmath0rb and molecular @xmath0rb@xmath55 bose - einstein condensates . in a conceptually related but experimentally quite different approach , thompson _ et al . _
@xcite have used an oscillating _
magnetic _ field to stimulate deexcitation of atom pairs from free atom states to molecular states of @xmath3rb@xmath55 .
1-photon and coherent 2-photon photoassociation can be viewed as manifestations of _ optical feshbach resonances _ , in which laser frequencies or intensities are used to tune collision properties . in a `` dressed state '' picture ,
the photons of a laser field bring atomic or molecular states at different energies into resonance with one another .
such resonances were first investigated theoretically by fedichev _
et al._@xcite , who showed that moderate laser intensities near resonance could induce sufficient excited - state character to change scattering lengths and even reverse their sign . bohn and
julienne @xcite extended this work and introduced the concept of 2-photon ( or 2-colour ) optical feshbach resonances , in which the state responsible for the resonance is a vibrational level of the _ ground _ electronic state , connected to the atomic state by a 2-photon ( stimulated raman ) transition .
optical feshbach resonances were first demonstrated experimentally by fatemi _
et al . _
@xcite , who observed changes in na - na scattering properties due to a one - photon resonance as a function of laser detuning and intensity .
et al._@xcite observed the variation of scattering length directly , by using bragg spectroscopy to determine the mean field energy of an @xmath0rb condensate , and thalhammer _
et al._@xcite have carried out similar measurements for a stimulated raman resonance .
an optical feshbach resonance can in principle be used to create molecules in a very similar way to a magnetic feshbach resonance .
javanainen and mackie @xcite proposed a photoassociation scheme in which molecules are produced coherently by chirping ( ramping ) the laser frequency adiabatically across an optical feshbach resonance .
et al . _
@xcite have investigated the formation of electronic ground - state @xmath0rb@xmath55 molecules in a similar way , and considered ramping the laser intensity as well as the frequency .
tuning through optical feshbach resonances is potentially more general than magnetic tuning .
laser fields can be switched on and off much faster than magnetic fields , and optical tuning could be applied to atoms without nuclear spin , such as the predominant isotopes of several of the alkaline earths ( @xmath109 mg , @xmath9ca , @xmath110sr , @xmath111ba ) .
in addition , levels can be tuned into resonance from much further away using 2-photon resonances than is possible magnetically . however
, a limitation arises because the extent to which the crossing is avoided depends on the laser - induced coupling between the two states , and thus on transition moments and franck - condon factors .
if the raman transition is very weak , it will be impossible to tune across the resonance slowly enough to achieve adiabatic passage .
most photoassociation experiments have so far used fixed - frequency lasers .
however , an alternative is to use laser pulses with tailored frequency and intensity profiles , as has become common in quantum control experiments on molecules at higher temperature @xcite .
a broadband laser can create a non - stationary state ( wavepacket ) that is made up of a linear combination of several different rovibrational levels of the excited electronic state .
the wavepacket then evolves in time , and if carefully chosen may develop favourable franck - condon overlap with low - lying vibrational levels of the electronic ground state .
vala _ et al . _
@xcite have simulated the use of chirped picosecond laser pulses to form cs@xmath55 molecules in the double - minimum @xmath105 state , and luc - koenig _
et al._@xcite have investigated optimization of the pulse characteristics .
et al._@xcite have simulated two - photon photoassociation using the scheme shown in fig .
[ chirp-2colour ] and optimized the parameters of the dump pulse to maximize the formation of molecules in deeply bound vibrational states .
et al . _
@xcite have carried out initial experimental work in which evolutionary strategies are used to optimize pulse parameters to maximise formation of ultracold @xmath3rb@xmath55 , while brown _ et al . _
@xcite have found that chirped femtosecond pulses produce _ fewer _ ultracold @xmath3rb@xmath55 and @xmath0rb@xmath55 molecules than comparable unchirped pulses .
[ htbp ] using chirped pump and dump pulses , showing the evolution of the wavepacket on the upper electronic state ( @xmath105 ) .
reprinted with permission from koch _
et al._@xcite .
copyright 2006 by the american physical society.,width=321 ] both photoassociation and magnetic resonance tuning produce molecules in very high vibrational states .
however , molecules in excited vibrational states can always undergo inelastic collisions that lead to trap loss .
there is thus great interest in finding ways either to drive the formation of molecules in the vibrational ground state , @xmath112 , or to transfer molecules initially formed in high - lying states to @xmath112 .
direct photoassociation to form low - lying vibrational states is not usually feasible for homonuclear molecules , because the low - energy scattering wavefunction for a pair of atoms has very little amplitude at short range .
there is therefore very little franck - condon overlap with the wavefunctions for low - lying vibrational states , which are entirely at short range .
various schemes have been proposed to overcome this @xcite , but for homonuclear molecules the combination of parity restrictions and franck - condon factors present formidable obstacles . for example ,
et al._@xcite proposed a 6-photon scheme to form @xmath0rb@xmath55 in its ground vibronic state .
for heteronuclear species , on the other hand , the parity restrictions are lifted and the franck - condon factors are more favourable @xcite , so that stimulated raman photoassociation to form @xmath112 molecules may be feasible @xcite .
[ htbp ] molecules by transferring them to the ground vibronic state . reprinted with permission from koch _
et al._@xcite .
copyright 2004 by the american physical society.,width=321 ] an alternative approach that may be advantageous is to form long - range molecules first , by either feshbach resonance tuning or photoassociation , and then transfer the molecular population to a short - range state .
et al . _
@xcite have used optimal control theory to design tailored laser pulses that would achieve this for na@xmath55 as shown in fig .
[ chirp - deep ] , while stwalley @xcite has suggested that for heteronuclear alkali metal dimers it could be efficiently achieved by stimulated raman adiabatic passage via mixed levels of the b@xmath113 and a@xmath114 states .
[ htbp ] in the culmination of a series of papers @xcite , sage _
et al._@xcite have succeeded in creating ultracold rbcs molecules ( @xmath115k ) in their vibronic ground state using the 2-step ( 4-photon ) process shown in fig .
[ rbcs ] : first , molecules are produced in the weakly bound @xmath116 level of the a@xmath117 state by 1-photon photoassociation ( a ) followed by spontaneous emission ( b ) , and then they are transferred to the @xmath112 or 1 level of the x@xmath114 state by an incoherent 2-photon pump / dump process ( stimulated emission pumping , sep , ( c ) and ( d ) in fig . [ rbcs ] ) via a mixed level of the @xmath118 and @xmath23 excited states .
the sep process has an efficiency of only 6% , but this could in principle be dramatically improved by using stirap instead .
other heteronuclear molecules such as krb @xcite and nacs @xcite have also been produced in the electronic ground state @xcite and state - selectively detected @xcite , but not yet transferred to low - lying vibrational states .
lics and nacs have been formed in the lowest vibrational level of the lowest triplet state on the surface of helium droplets @xcite , but the temperature is that of the droplet ( 0.38k ) and can not easily be lowered further .
an optical lattice is formed by standing waves between two or more laser beams . since atoms and molecules
are polarisable , they experience a periodic potential with minima at the points where the electric fields due to the laser are greatest .
the separation between successive minima is half the laser wavelength , and the heights of the barriers between minima can be adjusted by varying the laser intensity .
confinement of atoms in optical lattice cells was first observed by westbrook
_ et al._@xcite .
one particularly interesting state that can be created is a mott insulator phase @xcite , in which the lattice sites are occupied in a regular pattern and which can be melted " to form a superfluid by lowering the barriers .
the dynamics of bose - einstein condensates in optical lattices have been recently reviewed by morsch and oberthaler @xcite .
[ htbp ] the production of molecules in optical lattices offers intriguing possibilities .
the barriers between lattice sites enhance stability by preventing collisions between molecules at different sites .
et al . _
@xcite proposed creating a molecular bose - einstein condensate by raman photoassociation in a mott insulator with 2 atoms in each lattice site as shown in fig.[mol - lattice ] .
the trapping potential discretizes the molecular continuum and converts the free - bound photoassociation process into a bound - bound transition .
et al._@xcite extended this idea to the creation of a dipolar superfluid by photoassociation in a lattice containing one atom of each of two different species in each site .
et al._@xcite have created @xmath0rb@xmath55 molecules in an optical lattice by essentially the technique of ref .. ryu _ et al . _
@xcite have carried out similar experiments and observed coherent oscillations ( rabi cycling ) between atomic and molecular gases .
they also observed distinct raman spectra for photoassociation occurring in doubly and triply occupied sites .
et al._@xcite have also created @xmath0rb@xmath55 on an optical lattice , but by magnetic tuning through a feshbach resonance .
they were able to purify the resulting molecular lattice by driving out the remaining atoms with a laser , and showed that the molecules remained trapped much longer ( 700 ms ) after the atoms had been removed than while they were still present .
they attribute the loss rate to inelastic atom - molecule collisions that occur when atoms tunnel through lattice barriers into sites occupied by molecules ; once the atoms are removed , the corresponding loss due to molecule - molecule collisions is much slower simply because of the larger mass and reduced tunnelling rate of the molecules .
very recently , winkler _
et al._@xcite have created long - lived bound atom pairs with a _ repulsive _ atom - atom interaction .
the pairs can not decay into separated atoms because the energies of the atoms that would be produced are not allowed by the lattice band structure .
remarkably , the pairs _ do _ decay if the repulsive interaction between the atoms is switched off by tuning the scattering length to zero .
et al . _
@xcite have created a mott state of @xmath0rb@xmath55 molecules and demonstrated that phase coherence is restored when the lattice depth is reduced .
initial experiments have also been carried out on fermion dimers in optical lattices .
et al . _
@xcite reported the creation of a 1-dimensional gas of @xmath9k@xmath55 molecules in a 2-d optical lattice by feshbach resonance tuning , and stferle _ et al . _
@xcite have carried out similar experiments in a 3-d lattice . however , the tunnelling rates for the lattices used were too fast for lifetimes to be increased beyond those normal for fermion dimers .
optical lattices provide a very promising environment for bringing together more than two atoms at a time and studying many - body processes under controlled conditions .
stoll and khler @xcite have suggested a scheme that could be used to produce efimov states of alkali metal trimers directly from 3 atoms by magnetic feshbach resonance tuning in an optical lattice .
collision processes involving ultracold molecules are of prime importance to trapping and controlling them .
as described above , inelastic collisions usually release enough kinetic energy that both collision partners are lost from the trap .
the initial experiments on boson dimers formed by feshbach resonance tuning @xcite gave lifetimes that suggested vibrational relaxation rates for atom - molecule collisions around @xmath119 @xmath120 s@xmath121 .
quantitative estimations were not usually attempted , but this was consistent with the rate of @xmath122 @xmath120 s@xmath121 , estimated by yurovsky _
et al . _
@xcite on the basis of early feshbach resonance experiments @xcite .
it also agreed with quantum dynamics calculations by soldan _
et al . _
@xcite on vibrational relaxation in na + na@xmath55 collisions .
more recently , mukaiyama _ et al . _
@xcite have measured the trap loss rate for @xmath2na@xmath55 molecules formed by feshbach resonance tuning and obtained an atom - molecule rate coefficient @xmath123 @xmath120 s@xmath121 for molecules in the highest vibrational state .
relaxation processes involving molecules formed by photoassociation have also been studied .
et al._@xcite obtained an upper bound of @xmath124 @xmath120 s@xmath121 for @xmath0rb@xmath55 molecules in the second - to - last vibrationally excited state .
et al . _
@xcite have investigated inelastic collisions of rovibrationally excited cs@xmath55 ( @xmath125 ) in collisions with cs atoms in two different ranges of the vibrational quantum number @xmath126 by monitoring trap loss of cs@xmath55 .
they obtained atom - molecule rate coefficients close to @xmath127 @xmath120 s@xmath121 for both @xmath128 to 6 and @xmath129 to 47 .
et al . _
@xcite have carried out similar work for different rovibrational states of @xmath125 , but also considered molecules in the @xmath106 state and molecule - molecule collisions .
they obtained rate coefficients of @xmath130 @xmath120 s@xmath121 and @xmath131 @xmath120 s@xmath121 in the atom - atom and atom - molecule cases respectively , both with quite large error bounds .
soldn _ et al . _
@xcite , qumner _
et al . _
@xcite , and cvita _ et al . _
@xcite have carried out quantum dynamics calculations on atom - molecule collisions between alkali - metal atoms and dimers .
this work will be described in greater detail in a forthcoming review @xcite .
the calculations used a reactive scattering approach developed by launay and ledourneuf @xcite , which has been applied extensively to chemical reactions such as n(@xmath13d)+h@xmath55 @xcite and o(@xmath6d)+h@xmath55 @xcite at higher energies .
et al . _
@xcite showed that _ barrierless atom exchange reactions _ can occur in na + na@xmath132 , and that even at very low energy such collisions cause very fast vibrational relaxation ( @xmath133 on the order of @xmath119 @xmath120 s@xmath121 ) for collisions of molecules in low vibrational states .
the cross sections were shown to depend strongly on the details of the potential energy surfaces , and to change by a factor of 10 when non - additive forces were included , as shown in fig.[figna ] . in subsequent work ,
et al._@xcite showed that non - additive forces are important for all the alkali metal trimer systems , and affect the well depth for spin - polarized li + li@xmath55 collisions by a factor of 4 .
the quantum dynamics calculations were subsequently extended to li + li@xmath55 collisions , both isotopically homonuclear @xcite and heteronuclear @xcite , and to k + k@xmath132 @xcite .
for the homonuclear li systems , cvita _ et al . _
@xcite demonstrated that there is _ no _ systematic suppression of the inelastic collision rates for fermion dimers in low vibrational states , in contrast to the situation for molecules in long - range states @xcite .
this is very important for attempts to transfer molecules formed by feshbach resonance tuning to the vibrational ground state , because it means that the transfer must be accomplished _ without _ spending significant time in intermediate vibrational states .
there have been enormous advances in the production and manipulation of molecules in laser - cooled atomic gases .
molecules have been produced both by photoassociation and by magnetic tuning through feshbach resonances .
molecular bose - einstein condensates have been produced for molecules in long - range states , and the first signatures of triatomic and tetraatomic molecules have been seen .
most of the experimental advances were guided by theoretical predictions , and the experiments have in turn stimulated a large number of theoretical studies .
prospects for the future include the use of cold molecules for high - precision measurement and the production of quantum - degenerate gases of ground - state molecules , which will be stable to collisions and offer a wealth of new possibilities for quantum control .
heteronuclear molecules are particularly interesting , because they can have substantial dipole moments in long - range states .
dipolar quantum gases offer a new range of novel properties , and ultracold polar molecules also have potential applications in quantum computing and in studying fundamental physical properties such as parity violation and the electron dipole moment .
the authors are grateful to simon cornish and charles adams for valuable discussions and to eite tiesinga and cheng chin for providing figures based on unpublished work .
ps acknowledges support from the ministry of education of the czech republic ( grant no .
lc06002 ) .
|
this review describes recent experimental and theoretical advances in forming molecules in ultracold gases of trapped alkali metal atoms , both by magnetic tuning through feshbach resonances and by photoassociation .
molecular bose - einstein condensation of long - range states of both boson dimers and fermion dimers was achieved in 2002 - 3 .
condensates of boson dimers were found to be short - lived , but long - lived condensates of fermion dimers have been produced .
signatures of triatomic and tetraatomic molecules have recently been observed .
both homonuclear and heteronuclear molecules have been formed by photoassociation , mostly in very high vibrational levels .
recent attempts to produce ultracold molecules in short - range states ( low vibrational levels ) are described . experimental and theoretical work on collisions of ultracold molecules is discussed . to be published in international reviews in physical chemistry , vol .
25 , issue 4 ( oct - dec 2006 )
|
the observation of positive- and negative - parity states connected by e1 and e3 transitions in atomic nuclei is usually explained with the presence of quadrupole - octupole deformations @xcite . in the even - even nuclei one
typically observes alternating - parity bands , whereas in odd - mass nuclei the spectrum is characterized by a quasi parity - doublet structure @xcite .
the low - lying ( yrast ) structure of the quadrupole - octupole spectra was relatively well studied within different microscopic and collective model approaches @xcite@xcite ( see also @xcite and references therein ) .
however , the interpretation and the model classification of the higher , non - yrast parts of these spectra is still limited .
recently the model of coherent quadrupole - octupole motion ( cqom ) @xcite was applied to describe non - yrast collective bands with positive and negative parities in even - even nuclei together with attendant b(e1 ) , b(e2 ) and b(e3 ) transition probabilities @xcite .
it was shown that couples of @xmath1-bands and higher - energy ( non - yrast ) negative - parity bands can be interpreted within the model framework in the same way as the yrast alternating - parity bands . on this basis
it was concluded that the octupole degrees of freedom have a persistent role at higher energies and the quadrupole - octupole structure of the spectrum develops towards the non - yrast region of collective excitations @xcite .
the purpose of the present work is to examine the capability of the cqom model scheme to describe non - yrast quadrupole - octupole excitations in odd - mass nuclei by extending the original consideration proposed in @xcite .
for this reason it is assumed that some non - yrast level sequences with positive and negative parities observed in these nuclei can be associated to a higher - energy quasi parity - doublet structure of the spectrum .
it is supposed that such a structure can appear as the manifestation of higher - energy quadrupole - octupole rotation - vibration modes coupled to the single - particle motion .
the coupling of the odd nucleon to the even - even nuclear core as well as the coriolis interaction are taken into account phenomenologically . at the same time
the implemented approach is supposed to pave the way for a subsequent microscopic treatment of the odd - nucleon degrees of freedom . in this meaning the present work may be considered a necessary step towards a deeper understanding of the mechanism which governs the evolution of quadrupole - octupole dynamics in the higher - energy part of the spectrum in odd - mass nuclei .
this work is organized as follows . in sec .
[ sec:2 ] the cqom model formalism for the split parity - doublet bands and its features in the non - yrast part of the spectrum are briefly presented . in sec .
[ sec:3 ] numerical results and a discussion on the application of the model to a number of odd - mass nuclei in the rare - earth and actinide regions are presented . in sec .
[ sec:4 ] concluding remarks are given .
the model hamiltonian for odd - mass nuclei is taken as @xcite @xmath2 where @xmath3 and @xmath4 are axial quadrupole and octupole deformation variables , respectively , and @xmath5 here @xmath6 @xmath7 , @xmath8 @xmath9 and @xmath10 ( @xmath11 ) are quadrupole ( octupole ) mass , stiffness and inertia parameters , respectively . the quantity @xmath12 \ , \label{xdecoup}\end{aligned}\
] ] involves the total angular momentum @xmath13 , its third projection @xmath14 and the decoupling factor @xmath15 for the intrinsic states with @xmath16 .
the parameter @xmath17 determines the potential core at @xmath18 . in the present work
the decoupling factor is considered as a model parameter and is adjusted to the experimental data . in @xcite we show that a microscopically determined effect of the coriolis interaction can be included in cqom through an appropriate parity - projection particle - core coupling scheme in which the decoupling factor is calculated by using a reflection - asymmetric deformed shell model @xcite
. under the assumption of coherent quadrupole - octupole oscillations with a frequency @xmath19 and after introducing ellipsoidal coordinates @xmath20^{\frac{1}{2 } } \ \ \ \mbox{and}\ \ \ \phi=\arctan\left ( { \frac{\beta_3}{\beta_2 } \sqrt{\frac{d_3}{d_2}}}\right ) \ , \nonumber\end{aligned}\ ] ] such that @xmath21 with @xmath22 the collective energy of the system is obtained in the following analytic form @xcite @xmath23 , \label{enspect}\\ n&=&0,1,2, ... \ , \ \
k=1,2,3, ... \ , \nonumber\end{aligned}\ ] ] where @xmath24 has the meaning of a radial @xmath25 quantum number , @xmath26 corresponds to an angular @xmath27 quantum number and @xmath28 is considered as a parameter .
the quadrupole - octupole vibration wave function is @xmath29 where the radial part @xmath30 involves generalized laguerre polynomials in the variable @xmath31 with @xmath32 and @xmath33 .
the angular part in the variable @xmath34 appears with a positive or negative parity , @xmath35 , of the even - even core as follows @xmath36 the energy spectrum is determined in ( [ enspect ] ) by the quantum numbers @xmath24 and @xmath26 .
the parity - doublet structure is imposed by the condition @xmath37 , where @xmath38 is the parity of the odd - nucleon ( single - particle ) state .
then for a given state belonging to a parity - doublet the core parity is determined as @xmath39 .
the total core plus particle wave function for a state with an angular momentum @xmath40 belonging to a parity - doublet sequence in odd - even nuclei is given by @xcite @xmath41\ , \label{spwf}\end{aligned}\ ] ] where @xmath42 is the rotation wigner function and @xmath43 is the wave function of the odd - nucleon state . in the present work
it is supposed that the odd - nucleon parity @xmath38 is a good quantum number , although in the more general treatment of the single - particle motion in the octupole deformed ( reflection - asymmetric ) mean - field ( potential ) the single nucleon state may appear with a mixed parity @xcite-@xcite , @xcite .
then the parity - doublet is determined by a given @xmath24 and a pair of odd and even @xmath26-values , @xmath44 and @xmath45 ( @xmath46 ) , respectively .
the @xmath26-values are determined so that @xmath47 for @xmath48 and @xmath49 for @xmath50 .
the difference between @xmath51 and @xmath52 generates in ( [ enspect ] ) an energy splitting of the parity - doublet . that is why the term `` quasi parity - doublet '' was involved . according to the rule above the states having the same parity as the ground or bandhead state appear lower in energy and
are characterized by the odd @xmath51 number , while the opposite - parity states are shifted up and are labeled by an even ( and larger ) @xmath52 number .
the yrast doublet with @xmath53 is formed on the top of the ground state whose parity is @xmath54 .
the non - yrast doublets with @xmath55 are coupled to excited s.p . or quasi - particle ( q.p . ) states ( if the pairing correlations are taken into account )
whose parities @xmath56 determine the respective quasi - doublet structures according to the rule above .
also , the index @xmath24 labels the different intrinsic configurations to which the non - yrast doublets are coupled .
the above model mechanism of the forming of parity - doublet structures takes into account the possibility for a switch , at certain higher angular momentum , of the s.p .
state to which the doublet states are coupled to a state with an opposite parity , @xmath57 . as suggested in @xcite the switch
can be explained as the effect of an alignment process in the core @xcite . in the present model
this leads to a respective inversion of the parity @xmath58 of the vibration state at the given angular momentum and to a subsequent inversion of the mutual disposition of the parity - doublet counterparts , up - down @xmath59 down - up .
such a situation is indeed observed in few cases and described by the model ( see next section ) . in the original application of the model to the yrast octupole spectra
it was considered that for a given @xmath24 the nucleus always takes the lowest quadrupole - octupole vibration states with angular phonon numbers @xmath60 or @xmath61 depending on the parity @xcite .
the generalization of the model description by including non - yrast alternating - parity spectra of even - even nuclei showed that a better agreement with the experimental data can be obtained if higher @xmath62-values are also allowed @xcite .
it was seen that the difference @xmath63 reasonably specifies the mutual disposition of the opposite - parity sequences in the spectrum .
therefore , it can be expected that the same meaning of the higher @xmath26-values will be valid for the parity - shift in the quasi - doublet bands of odd - mass nuclei . on the other hand , as recognized in @xcite , in this case one has jumps of the quantum numbers @xmath64 and @xmath65 over several low - lying angular - phonon excitations within the set of levels characterized by given radial - phonon number @xmath24 .
presently this is only justified by the meaning of @xmath66 as a characteristics of the mutual displacement of the opposite - parity sequences .
the search for a deeper meaning and more sophisticated correlation between the quadrupole and octupole modes capable to compensate or explain these jumps is still an open issue . in this work
we apply the original concept for the lowest @xmath67 or @xmath61 phonon numbers in the model description of yrast and non - yrast quasi parity - doublet spectra of odd-@xmath0 nuclei . at the same time in the end of the next section we discuss the difference between the obtained results and the result of calculations performed for the same nuclei by allowing higher @xmath65 values for the upper shifted doublet counterparts and fixed @xmath68 value for the counterparts whose parity coincides with the bandhead parity . for this reason we keep the formalism in its general form capable of treating unrestricted values of angular - phonon numbers . by using the wave functions ( [ spwf ] )
one can calculate b(e@xmath69 ) transition probabilities ( @xmath70 ) in the yrast and non - yrast quasi - doublet spectra .
the relevant formalism was originally developed in @xcite and further extended to the non - yrast states of even - even nuclei @xcite . in this work
we apply the formalism developed in @xcite . due to the imposed axial symmetry the b(e@xmath69 )
probabilities are non - zero only between states with the same @xmath14-values .
also , here the odd - nucleon wave function , @xmath43 in ( [ spwf ] ) , is not considered explicitly , while its parity is taken into account through the above explained parity - coupling scheme . as a result in this work we consider transition probabilities between states belonging to the same yrast or non - yrast quasi parity - doublet and , therefore , coupled to the same s.p .
state ( doublet bandhead ) .
then the cqom reduced electric transition probabilities b(@xmath71 ) with multipolarity @xmath70 between the initial ( i ) and final ( f ) doublet states have the form @xmath72 where @xmath73 corresponds to the same s.p .
state . the factors @xmath74
involve integrals over the radial @xmath31 and the angular @xmath34 variables and can be written in the following compact form @xmath75 with the exponents @xmath76 , @xmath77 and indexes @xmath78 for @xmath70 , respectively .
the integrals over @xmath31 are @xmath79 while the integrals over @xmath34 read @xmath80 where the factors @xmath81 represent series expansions defined in eqs .
( 28)-(30 ) of @xcite .
analytic expressions for the @xmath31-integrals ( [ si ] ) and explicit expressions for the @xmath34-integrals are given in appendices b and c of @xcite , respectively .
the quantities @xmath74 depend on the multipole charge factors @xmath82 , as determined in eqs .
( 22 ) and ( 23 ) of @xcite .
the parameters @xmath83 and @xmath84 are defined in eq .
( [ pqd ] ) above .
however as shown in @xcite they are not independent but one has @xmath85 , so that only @xmath83 is kept as an adjustable parameter .
also , the integrals ( [ si ] ) depend on the parameter @xmath86 which enters the radial wave functions ( [ psieta1 ] ) . in addition , an effective charge @xmath87 is introduced to determine the correct scale of the b(e1 ) transition probabilities with respect to b(e2 ) @xcite .
the consideration of transition probabilities between states belonging to different parity - doublets , and with @xmath88 , can be implemented after obtaining @xmath43 within the reflection - asymmetric deformed shell model ( dsm ) @xcite as done in @xcite and by taking into account the coriolis @xmath14-mixing effect in the respective s.p .
states as proposed in @xcite .
such a task implies a natural connection of the collective cqom model to an approach in which the ground state , the excited q.p . states , the respective decoupling factors ( for @xmath16 ) , as well as the coriolis mixing contributions are obtained microscopically .
although the work in this direction is in an essential progress @xcite , here we consider the s.p .
degree of freedom phenomenologically .
we consider that the knowledge of the pure collective cqom description of quadrupole - octupole spectra , especially the yrast and non - yrast quasi - doublet levels , is a necessary step before attempting detailed implementation of the microscopic part of the approach .
thus for a given doublet we take the s.p .
parity and the @xmath14-value as suggested by the experimental analysis or by microscopic calculations reported in the literature .
the quasi - doublet band - heads are obtained as different rotation - vibration modes characterized by the cqom oscillator quantum number @xmath89 . in the cases of @xmath16
the decoupling factors @xmath90 for the respective doublets , labeled by @xmath24 and entering the expression ( [ xdecoup ] ) , are adjusted according to the experimental data .
it should be noted that these phenomenological decoupling factors are of special importance for determining the physically reasonable deformation regions where dsm calculations have to be performed after inserting the microscopic part in the cqom .
( see @xcite for more details on this consideration . )
in this section results of the cqom model calculations for several odd - mass nuclei are presented .
the model energy levels are determined by eq .
( [ enspect ] ) as @xmath91 where @xmath92 , @xmath93 , @xmath94 , @xmath95 and @xmath96 correspond to the ground state .
the model parameters @xmath97 , @xmath98 and @xmath99 determine the energy levels , while the parameters @xmath86 , @xmath83 and the effective charge @xmath87 determine in addition the transition probabilities as explained below eq .
( [ ilam ] ) .
these parameters are adjusted by simultaneously taking into account experimental data on the energy bands @xcite and the available b(e1 ) and
b(e2 ) transition probabilities @xcite .
( data on b(e3 ) probabilities are not available . ) the theoretical b(e1 ) and b(e2 ) values are calculated through eq .
( [ bela ] ) . in the case(s ) of @xmath16 bandheads the decoupling factor(s
) @xmath90 ( @xmath53 for the yrast , and @xmath100 for the first non - yrast doublets ) is ( are ) also adjusted .
calculations were performed for the nuclei @xmath101pm , @xmath102gd , @xmath103ra , @xmath104np and @xmath105am . for each of them the yrast and one non - yrast quasi parity - doublet
are considered together with a number of experimentally observed b(e1 ) and b(e2 ) transition probabilities .
the theoretical and experimental energy levels of the nuclei @xmath101pm , @xmath102gd and @xmath103ra are compared in figs .
[ f1 ] and [ f2 ] . in fig .
[ f3 ] both the energy levels and the transition probabilities for the nuclei @xmath104np and @xmath105am are given .
the obtained parameter values are given in the figures . also , the root mean square ( rms ) deviations between the theoretical and experimental levels of the yrast band ( rms@xmath106 ) , the first non - yrast band ( rms@xmath107 ) and the total rms factors for both bands ( rms@xmath108 ) are given in figs .
[ f1]-[f3 ] .
the theoretical and experimental b(e1 ) and b(e2 ) transition probabilities for all considered nuclei are compared in table 1 . for each nucleus
the model classification scheme accommodates one excited non - yrast quasi parity - doublet band . in all
nuclei the theoretical energy sequences reproduce the structure of the experimentally observed yrast and non - yrast bands .
this makes it meaningful to predict a few more states in some of the considered bands in order to show how the respective doublet - structures could develop with the angular momentum .
more specifically , in figs .
[ f1]-[f3 ] we add a number of predicted states to one of the parity - counterpart sequences in order to get an equal number of positive- and negative - parity levels in the theoretically obtained quasi - doublets .
the quasi - doublet sequences in the nuclei @xmath101pm and @xmath102gd given in fig .
[ f1 ] have similar overall structures . in @xmath101pm
we have a relatively well developed yrast doublet built on a @xmath109 ground state and a shorter non - yrast doublet considered to be coupled to a @xmath110 s.p . state . in the latter a strong coriolis decoupling effect is observed , such that some higher - spin states appear at lower energy .
this spin - interchange effect is partly reproduced in the model description through a relatively large decoupling - parameter value @xmath111 . at the same time in the experimental yrast sequence
we indicate an inversion of the up - down shift between the opposite - parity counterparts at @xmath112 . this effect corresponds , according to the model assumption explained in the previous section , to an inversion of the s.p .
parity and imposes an interchange of the @xmath113 and @xmath114 values in the model spectrum for @xmath115 . as a result the observed shift of the positive - parity states above their negative - parity counterparts is reproduced . in @xmath102gd
both quasi - doublets are built on negative - parity s.p .
states with @xmath116 which leads to the shift of the positive - parity counterparts in the lower part of the spectrum up in energy compared to the negative - parity ones .
this is described according to the model scheme by assigning @xmath117 to the positive - parity states and @xmath68 ( @xmath118 ) to the negative - parity ones . similarly to @xmath101pm , in the yrast sequence of @xmath102gd the two @xmath119 values are interchanged at @xmath120 .
we remark that for both nuclei in fig .
[ f1 ] the two @xmath26-values given above the theoretical yrast - band sequences correspond to the respective positive- or negative - parity counterparts before the interchange . in @xmath103ra , given in fig .
[ f2 ] , the yrast band is described by imposing @xmath110 instead of the experimentally suggested @xmath121 .
this is motivated by the previously indicated staggering behavior of the doublet splitting which is considered a manifestation of a coriolis decoupling effect in that band @xcite .
we remark that the decoupling factor obtains the same value @xmath122 as in the model description obtained in @xcite for the yrast band only .
3(b ) of @xcite this value was misprinted without the correct sign @xmath123 . )
the considered non - yrast quasi - doublet is built on a @xmath109 bandhead state .
we see that the overall disposition of this doublet with respect to the yrast band is reasonably reproduced .
the parity splitting in the yrast sequence is a bit overestimated by the theory , while that in the non - yrast band is slightly underestimated .
nevertheless , the root mean square ( rms ) deviations between the theoretical and experimental levels of the different bands as well as the total rms factor are obtained in quite reasonable limits between 20 and 25 kev ( see fig . [ f2 ] ) . the description of @xmath104np and @xmath105am given in fig .
3 illustrates the applicability of the model scheme in the region of heavier nuclei .
here we can remark that the strong coriolis effect in the non - yrast band of @xmath104np is adequately described with a decoupling factor @xmath124 .
as a result the situating of the @xmath125 state below the @xmath126 state and the @xmath127 state below the @xmath128 state in this band is well reproduced . at the same time
the model predicts that in the positive - parity sequence of the non - yrast band the @xmath129 state will be placed below the @xmath130 state .
the theoretical values of the b(e1 ) and b(e2 ) transition probabilities , given in table i , show an overall good agreement between theory and experiment .
this makes it meaningful to predict a few more transition - probability values , especially between the lowest opposite - parity states in the non - yrast doublets ( e1 transitions ) as well as between states within sequences with the same parity ( e2 transitions ) .
such predictions are given for all considered nuclei .
it should be noted that data on transition probabilities in the non - yrast parts of the considered spectra are not available .
in addition , we should remind that transitions between states belonging to different quasi parity - doublets are not included in the present consideration . for a number of yrast - band transitions
larger discrepancies between theory and experiment are observed . in some cases , such as the e1 transitions in @xmath103ra ,
this can be explained with the more complicated structure due to the presence of the coriolis interaction .
however , we remark on the good model reproduction of the experimental b(e1 ) transition values in @xmath101pm , @xmath104np and @xmath105am . in the nuclei @xmath104np and @xmath105am for which only one b(e2 ) value is described
we have an exact agreement between the theory and experiment due to the exact determination of the parameters @xmath86 and @xmath83 in the fitting procedure . cccc mult & transition & th [ w.u . ] & exp [ w.u . ]
+ + e1&@xmath131 @xmath132 @xmath133 & 0.0011&0.0014 ( 2 ) + e1&@xmath134 @xmath132 @xmath133 & @xmath135 & @xmath136 + e1&@xmath134 @xmath132 @xmath137 & @xmath138 & @xmath139 + e2&@xmath134
@xmath132 @xmath131 & 8 & @xmath140 + e2&@xmath137 @xmath132 @xmath133 & 85 & + e1&@xmath141 @xmath132 @xmath142 & @xmath143 & + e1&@xmath144 @xmath132 @xmath145 & @xmath146 & + e2&@xmath145 @xmath132 @xmath142 & 51 & + + e1&@xmath133 @xmath132 @xmath147 & @xmath148 & @xmath149 ( 8) + e1&@xmath133 @xmath132 @xmath131 & @xmath150 & @xmath151 ( 22 ) + e2&@xmath131 @xmath132 @xmath147 & 311 & 293 + e2&@xmath134 @xmath132 @xmath147 & 130 & 119 + e2&@xmath134 @xmath132 @xmath131 & 195 & 230 + e2&@xmath137 @xmath132 @xmath133 & 18 & + e1&@xmath152 @xmath132 @xmath144 & @xmath153 & + e1&@xmath154 @xmath132 @xmath144 & @xmath155 & + e2&@xmath152 @xmath132 @xmath145 & 30 & + e2&@xmath156 @xmath132 @xmath144 & 199 & + + e1&@xmath147 @xmath132 @xmath157 & @xmath158 & @xmath159 ( 16 ) + e1&@xmath147 @xmath132 @xmath133 & @xmath160 & @xmath161 ( 9 ) + e1&@xmath134 @xmath132 @xmath133 & @xmath162&@xmath163 ( 24 ) + e1&@xmath134 @xmath132 @xmath137 & @xmath164 & @xmath165 ( 22,-5 ) + e2&@xmath134 @xmath132 @xmath147 & 17 & 10 ( 6 ) + e2&@xmath137 @xmath132 @xmath133 & 18 & 70 + e2&@xmath137 @xmath132 @xmath157 & 148 & 44 + e2&@xmath166 @xmath132 @xmath137 & 210 & 280 ( 12 ) + e1&@xmath144 @xmath132 @xmath152 & @xmath167 & + e1&@xmath144 @xmath132 @xmath154 & @xmath168 & + e1&@xmath156 @xmath132 @xmath154 & @xmath169 & + e2&@xmath154 @xmath132 @xmath152 & 298 & + e2&@xmath156 @xmath132 @xmath144 & 32 & + + e1&@xmath131 @xmath132 @xmath133 & @xmath170 & @xmath171 ( 11 ) + e1&@xmath131 @xmath132 @xmath137 & @xmath172 & @xmath173 ( 4 ) + e1&@xmath134 @xmath132 @xmath133 & @xmath174 & @xmath175 + e1&@xmath134 @xmath132 @xmath137 & @xmath176 & @xmath177 + e2&@xmath134 @xmath132 @xmath131 & 600 & @xmath178 + e2&@xmath179 @xmath132 @xmath134 & 562 & + e1&@xmath180 @xmath132 @xmath142 & @xmath181 & + e1&@xmath141 @xmath132 @xmath142 & @xmath181 & + e2&@xmath180 @xmath132 @xmath141 & 915 & + + e1&@xmath133 @xmath132 @xmath131 & @xmath173 & @xmath182(4 ) + e1&@xmath133 @xmath132 @xmath134 & @xmath158 & @xmath183(3 ) + e2&@xmath134 @xmath132 @xmath131 & 574 & 574 ( 9 ) + e2&@xmath137 @xmath132 @xmath133 & 61 & + e1&@xmath145 @xmath132 @xmath141 & @xmath184 & + e1&@xmath145 @xmath132 @xmath144 & @xmath185 & + e1&@xmath152 @xmath132 @xmath144 & @xmath186 & + e1&@xmath154 @xmath132 @xmath156 & @xmath187 & + e2&@xmath144 @xmath132 @xmath141 & 767 & + e2&@xmath152 @xmath132 @xmath145 & 79 & + here we can summarize that in all considered nuclei the rms deviations between the theory and experiment obtained for the different sequences as well as the total rms factors prove the good quality of the model description .
we remark that the total rms factors do not exceed 50 kev , with the lowest rms@xmath108=23 kev being obtained for @xmath103ra and the higher one , 48 kev , observed in @xmath101pm .
as mentioned in the previous section it can be expected that considerably better descriptions with lower rms factors may be expected if higher @xmath188 values are allowed for the upper quasi - doublet counterparts .
we have performed such calculations for the same data in the same nuclei . for each nucleus
the calculations were performed on a net over @xmath52 @xmath189 varying between @xmath61 and @xmath190 while @xmath51 was kept equal to 1 . in this way
the @xmath52 values which provide the best overall description of energy levels and transition probabilities were determined . as a result
, essentially lower energy rms factors were obtained in : @xmath103ra with `` favored '' @xmath191 and @xmath192 values and rms@xmath108=7 kev ; @xmath104np with @xmath191 and @xmath193 and rms@xmath194=15 kev ; @xmath105am with @xmath191 and @xmath192 with rms@xmath108=18 kev .
it is seen that in these nuclei the rms@xmath108 values obtained at @xmath188 are smaller than the values ( given in figs . 2 and 3 ) obtained for @xmath195 by a factor between 2 and 3 .
for the nuclei @xmath101pm with `` favored '' @xmath196 and @xmath193 values and rms@xmath194=37 kev , and @xmath102gd with @xmath197 and @xmath192 and rms@xmath108=29 kev the improvement of the description is not too pronounced ( see the rms values in fig . [ f1 ] ) .
the conclusion from the above mentioned calculations is that indeed the involvement of larger numbers of angular phonons leads to better model descriptions .
nevertheless the result illustrated in figs .
[ f1]-[f3 ] shows that the imposing of the lowest quadrupole - octupole vibration modes with @xmath68 and @xmath117 still provides quite adequate interpretation and classification of the yrast and first non - yrast quasi - doublet energy sequences in the considered odd - mass nuclei .
this means that the cqom model is capable of reproducing the specific spectroscopic properties related to the simultaneous manifestation of quadrupole and octupole degrees of freedom in these nuclei .
the use of the model with larger angular - phonon numbers can extend its applicability in wider regions of nuclei and quasi - doublet spectra , including higher non - yrast bands , but needs a detailed justification of the presence of lower model states for which experimental data are not observed .
it should be noted that even on the present level of phenomenology , the cqom model description provides a useful basis for further developments . as it was discussed in the end of sec .
[ sec:2 ] the knowledge of the decoupling factors as well as the model mechanism for the forming of parity - doublet spectra in odd - mass nuclei can guide the inclusion of microscopic calculations in the study .
in conclusion , the present work reports an application of the collective model of coherent quadrupole and octupole motion ( cqom ) for the description of yrast and non - yrast quasi - doublet spectra in odd - mass nuclei .
the calculations are performed by considering a zero number of radial quadrupole - octupole phonons in the yrast band and one radial - phonon excitation in the first non - yrast band . in both bands
the lowest possible angular - phonon modes in the motion of the even - even core are considered .
the results obtained for a number of odd - mass nuclei in the rare - earth and actinide regions illustrate the capability of the model to reproduce the structure of yrast and non - yrast energy levels together with the attendant b(e1 ) and b(e2 ) transition probabilities .
the test of the model by allowing a higher number of angular phonons shows a better agreement with the experimental data , but still needs a justification of the jumps over the lower phonon numbers . at the same time the collective rotation - vibration structure of the spectra and the observed coriolis decoupling effects are adequately taken into account . on this basis
the present cqom model descriptions can serve as a starting point for the application of a deeper collective and microscopic approach in the exploration of nuclear quadrupole - octupole collectivity .
work in this direction is in progress .
|
the model of coherent quadrupole and octupole motion ( cqom ) is applied to describe non - yrast split parity - doublet spectra in odd - mass nuclei .
the yrast levels are described as low - energy rotation - vibration modes coupled to the ground single - particle ( s.p . )
state , while the non - yrast parity - doublet structures are obtained as higher - energy rotation - vibration modes coupled to excited s.p . states .
it is shown that the extended model scheme describes both the yrast and non - yrast quasi parity - doublet spectra and the related b(e1 ) and b(e2 ) transition rates in different regions of heavy odd-@xmath0 nuclei .
the involvement of the reflection - asymmetric deformed shell model to describe the single - particle motion and the coriolis interaction on a deeper level is discussed .
|
it is well known that the resonant tunneling through a localized level results in a quantized conductance @xcite , which was observed in the transport through quantum dots @xcite .
recently , researchers @xcite found that the resonant tunneling with a quantized conductance also happens in the presence of majorana bound states , providing a way of detecting the majorana fermions @xcite . from then on , there is a growing interest in the study of the quantum transport through the systems that accommodate majorana bound states @xcite . the majorana fermion is a proposed charge - neutral particle of its own antiparticle @xcite , which continuously attracts the efforts of searching for its existence in nature @xcite , due to its potential application in topological quantum computation @xcite
the majorana fermions were predicted to emerge as the edge quasi - particle excitations of the topological superconductors @xcite .
some proposals have been introduced for observing the majorana fermions in these systems , e.g. , by the topological josephson effect ( see ref .
@xcite for recent progress ) . but no experiments realized these proposals up to now .
et al . _
@xcite proposed a method for detecting the majorana bound states in topological superconductors by the effect of the resonant andreev reflection across a junction between a normal metal and a topological superconductor , which contributes a quantized conductance of @xmath5 .
recent experimental results @xcite are qualitatively consistent with the existence of majorana bound states in these systems .
however , the observed zero bias peak in experiments does not exclusively signal the majorana bound states @xcite , mainly because the experiments were conducted in real materials where the disorder is unavoidable and may induce additional zero bias peak . to better understand the experimental results , it is necessary to study the effect of disorder @xcite on the transport through topological superconductors .
compared to the random disorder , the disorder produced by incommensurate potentials can induce a finite transition from extended states to anderson localized states in one - dimensional incommensurate lattices @xcite . due to its good controllability , the incommensurate potential has been experimentally engineered with ultracold atoms loaded in 1d bichromatic optical lattices @xcite , simulating intensive study of the localization properties of quasi - periodic systems @xcite .
particularly , the effect of incommensurate potentials on the one - dimensional topological superconductors has also been theoretically studied @xcite . as a prototype model for describing the one - dimensional topological superconductors ,
the kitaev s p - wave superconductor model has been widely used as an effective model in current theoretical studies from various aspects @xcite .
its realization in cold atom quantum wires is discussed by jiang _
et al . _
furthermore , fulga _ et al . _
@xcite discussed the possibility of realizing the kitaev chain in experiments by a chain of quantum dots coupled to superconductors .
the incommensurate potentials can be produced by well tuning the gate voltages applied to different dots .
another different proposal is by using the quantum wires @xcite , while the incommensurate potentials can be generated electrically @xcite .
the kitaev s model @xcite describes a p - wave topological superconductor .
the recent study @xcite discovered that in the presence of incommensurate potentials the system undergoes a quantum phase transition from the superconducting phase to the anderson localized phase as the strength of disorder increasing .
this conclusion is supported by the energy spectra and the calculation of the topological invariant .
however , the study of the transport through a junction containing the kitaev chain with incommensurate potentials is still lack . in this paper
, we employ the numerical operator method @xcite to study this problem .
our results clarify the effect of incommensurate disorder on the transport through topological superconductors , and provide information of distinguishing the different phases by the feature of the current - voltage curves .
furthermore , the study of transport theory and experiment for transition from the metal to the localized state in one - dimensional setups has never been carried out before since the traditional one - dimensional anderson model has no a finite transition .
the transport analysis in our paper fills this gap .
the rest of paper is organized as follows . in sec .
ii , we briefly review the kitaev s p - wave superconductor model with incommensurate potentials , introduce our junction model for studying the transport , and discuss the details of the numerical operator method .
we present the results of the current - voltage curves and the linear conductance as the gate voltage is zero in sec .
iii , and those as the gate voltage deviates from zero in sec .
v is a short summary .
the hamiltonian describing the kitaev chain with incommensurate potentials is written as @xmath6 where @xmath7 denotes the length of the chain , @xmath4 the hopping amplitude , @xmath3 the p - wave pairing amplitude induced by a rf field in cold atom systems @xcite or by the proximity effect in solid state systems @xcite , and @xmath8 the incommensurate potential varying at each lattice site . in the expression of @xmath9 , @xmath10 denotes the gate voltage or the overall shift of the energy levels , @xmath11 the strength of disorder , and @xmath12 an irrational number taken to be @xmath13 in the present study .
this hamiltonian @xcite describes a p - wave topological superconductor supporting zero - energy majorana bound states localized at the edges of the chain for @xmath14 . as @xmath15
the system experiences a quantum phase transition @xcite from the topological superconducting phase to the anderson localized phase .
the hamiltonian ( [ hamiltoniansc ] ) can be diagonalized as @xmath16 by introducing a bogoliubov - de gennes ( bdg ) transformation @xmath17 , where @xmath18 denotes the quasi - particle energy , and @xmath19 and @xmath20 can be obtained by numerically solving the corresponding bdg equations . in order to obtain the tunneling current in the following , we need to calculate the ground - state correlation functions @xmath21 ,
@xmath22 , @xmath23 and @xmath24 ( see the appendix [ appendix : corr ] for more details ) , written in terms of the amplitude of the bdg transformation as @xmath25 to study the transport through the kitaev chain with incommensurate potentials , we locate it between two single - channel normal leads ( the n - s - n structure has been used with random disorder @xcite ) , described by the hamiltonians @xmath26 and @xmath27 respectively , where @xmath28 is the hopping amplitude in the leads .
the leads @xmath7 and @xmath29 are coupled to the left and right edge sites of the kitaev chain respectively ( see fig . [
fig : schematic ] for the schematic diagram of the setup ) .
the hamiltonian describing the coupling between the leads and the chain is expressed as @xmath30 where @xmath31 is the coupling strength .
then the total hamiltonian of the transport setup is @xmath32 where @xmath33 is described by eq .
( [ hamiltoniansc ] ) .
we study the tunneling current from the left lead to the chain , expressed as @xmath34 and that from the chain to the right lead , expressed as @xmath35 in this paper , the two leads are in different chemical potentials , being @xmath36 and @xmath37 respectively , where @xmath38 is the voltage bias .
we take the wide band limit in the leads and the intermediate coupling strength between the leads and the chain by setting @xmath39 and @xmath40 .
the numerical operator method @xcite is employed to calculate @xmath41 and @xmath42 . for simplicity
, we define the synonyms @xmath43 and @xmath44 in this subsection .
we first solve the heisenberg equation @xmath45 $ ] to obtain the expressions of @xmath46 , @xmath47 , @xmath48 and @xmath49 .
since @xmath50 is quadratic , the solution is supposed to be @xmath51 analytically solving the propagators @xmath52 is difficult in the presence of disorder , however , they can be obtained through the numerical calculation iteratively . from @xmath53 and eq .
( [ heisenbergsolution ] ) , we have @xmath54 \\ & + \frac{(i\delta t)^2}{2 } [ \hat h,[\hat h,\hat d_{ks ' } ] ] + o(\delta
t^3 ) \bigg ) , \end{split}\ ] ] where the time step @xmath55 is set to be small enough so that the terms @xmath56 can be neglected .
@xmath57 can then be expressed as a linear function of @xmath52 : @xmath58 where the coefficients @xmath59 are defined by the commutator @xmath60= \sum_{ls_2 } g_{ks_1,ls_2}\hat d_{ls_2 } $ ] .
we work out the propagator @xmath52 at an arbitrary time by an iterative algorithm starting from @xmath61 when @xmath62 and moving forward @xmath55 at each step .
the error caused by a finite @xmath55 is in the order @xmath56 in our calculation .
one can keep the higher order terms @xmath63 with @xmath64 in eq .
( [ iterativerelation ] ) so as to reduce the error to the order @xmath65 .
this makes it possible to use a larger @xmath55 without reducing the precision of the result .
but , in practice , we find that keeping to the order @xmath66 is efficient enough for our purpose . at each step
, we store all the non - zero propagators @xmath52 and use them to calculate the propagators at next step .
the number of non - zero propagators increases with the step quickly .
so we apply a truncation scheme to keep only a fixed number of non - zero propagators ( the number is denoted by @xmath67 ) with the largest magnitudes at each step .
this truncation scheme is necessary for keeping the computing resource within a reasonable amount .
the value of @xmath67 is determined empirically . in this paper ,
setting @xmath67 to several tens of thousands is enough for obtaining the current with the relative error below @xmath68 .
then we substitute the solution in eq .
( [ heisenbergsolution ] ) into eqs .
( [ leftcdef ] ) and ( [ rightcdef ] ) , and express the currents as @xmath69 \langle \hat d_{js } \hat d_{j 's ' } \rangle , \\
i_r(t ) & = -2 g_c \sum_{j , s , j',s'}\textbf{im } \left [ w_{l-1,1,js}(t ) w_{l0,j's'}(t)\right ] \langle \hat d_{js } \hat d_{j 's ' } \rangle . \end{split}\ ] ] here the correlation function is @xmath70 as @xmath71 . and as @xmath72 and @xmath73 are both in the lead @xmath7 or @xmath29 , we have @xmath74 with @xmath75 as @xmath76 or @xmath77 as @xmath78 . the correlation function is zero in all the other cases . to demonstrate the power of our method
, we compare it with exact diagonalization at @xmath79 ( see more details in appendix [ appendix : comp ] ) .
the comparison shows that , for the long relaxation time of currents , our method gives better performance than the exact diagonalization method . .
the p - wave pairing amplitude and the voltage bias are set to @xmath80 .
we compare the left and right currents @xmath41 and @xmath42 at different disorder strength @xmath11 and gate voltage @xmath10 , represented by different types of points .
the solid lines are the guide for the eyes .
the left and right currents coincide well with each other except for @xmath81 and @xmath82 .
[ inset ] the @xmath1 curve obtained at @xmath83 and @xmath84 , in which case there is a single @xmath1 curve since @xmath85 . ]
what are interesting in experiments are the stationary currents @xmath86 and @xmath87 . in practice , we obtain @xmath88 and @xmath89 from @xmath41 and @xmath42 respectively at the times much larger than the current relaxation time . as shown in fig .
[ fig : evolution ] , @xmath41 and @xmath42 quickly relax to their stationary values within a few oscillations in the absence of disorder . while at a large disorder strength , the current relaxes slowly , since the disorder blocks the movement of electrons ( see the curves titled @xmath81 in fig .
[ fig : evolution ] ) .
we obtain the stationary currents @xmath88 and @xmath89 at different voltage biases and plot the corresponding current - voltage ( @xmath1 ) curves ( see the inset of fig . [ fig : evolution ] for an example ) .
the linear conductance is the slope of the @xmath1 curve at @xmath90 , and in practice is got by calculating the ratio of the current to the voltage bias at small @xmath38 in the linear response regime : @xmath91 where @xmath92 .
the conductance @xmath93 at @xmath94 as a function of the disorder strength @xmath11 , read out from the @xmath1 curves ( for the cases of given @xmath11 see fig .
[ fig : ivcurve ] ) .
the p - wave pairing amplitude is @xmath83 and the gate voltage is @xmath82 .
this conductance is just the linear conductance except for the points close to the critical point @xmath95 , where the linear response regime is smaller than @xmath94 .
[ bottom panel ] energy spectra of the kitaev chain at the same parameters under the open boundary condition .
only the quasi - particle energies @xmath96 are shown due to the particle - hole symmetry .
the energy gap closes at the point @xmath95 ( not clearly seen in the figure because the gap close to the transition point is too small ) , coinciding with the result in ref .
@xcite ] as @xmath97 , the kitaev chain is in the superconducting phase .
the voltage drops only at the interfaces between the chain and the leads , but not inside the chain .
the chemical potential of the chain is fixed to @xmath98 , i.e. , the chain is grounded .
the pair of majorana bound states locates exactly at zero energy , inside the transmission window @xmath99 $ ] .
so we observe the linear conductance of @xmath0 independent to the disorder strength @xmath11 ( see the top panel of fig . [
fig : transition ] ) , the same conductance as that of the resonant tunneling through a single level without the spin degrees of freedom .
the linear conductance disregarding the disorder is protected by the superconducting gap , which isolated the majorana bound states from those in the bands ( see the energy spectra in the bottom panel of fig . [
fig : transition ] ) . in the transmission window , there are only a pair of majorana states left , which locate at the two edges of the chain , coupled strongly to the left and right leads .
the majorana states participate in shuttling the electrons , resulting in the quantized conductance , which signals the resonant tunneling via the majorana bound states .
law et al . predicted in 2009 that the linear conductance of the junction between a normal lead and a topological superconductor is @xmath5 due to the resonant andreev reflection @xcite .
our result coincides exactly with theirs .
the current goes across two junctions in our model , each of which contributes a resistance of @xmath100 according to law et al .
, so that the total conductance of the two parallel - connected junctions is exactly @xmath0 .
furthermore , a quantized conductance excludes the possibility of the chain s being in the normal metal phase . in fig .
[ fig : ivcurve ] , we show the current - voltage curves of the system in different phases .
the linear conductance is the slope of the @xmath1 curve at the origin .
we see that the linear conductance of the normal metal phase ( @xmath101 ) is obviously lower than that of the topological superconducting phase ( @xmath102 ) .
even a chain in the normal metal phase also has scattering states around zero energy , but these states are not localized at the edges .
in fact , the coupling strength of these states to the leads scales as @xmath103 where @xmath7 denotes the chain length , becoming infinitesimal in the thermodynamic limit , while , the majorana states locate always at the edges and couple strongly to the leads , independent to the size of the chain .
so the resonant tunneling can happen via the majorana states , but not via the scattering states of the normal metal phase .
curves in the clean topological superconducting phase at @xmath104 ( daggers ) , in the disordered topological superconducting phase at @xmath105 with @xmath88 ( full dots ) and @xmath89 ( full squares ) separately representing the left and right currents , in the normal metal phase at @xmath106 ( crosses ) , and in the anderson localized phase at @xmath107 ( stars ) . at @xmath105
, we also rotate the chain by @xmath108 and reconnect it to the leads , and plot the corresponding left ( empty circles ) and right currents ( empty squares ) respectively .
the solid lines are the guide for the eyes .
the left and right currents are the same , except for @xmath105 . at @xmath105 ,
the left and right currents exchange with each other as we rotate the chain by @xmath108 , which is shown by the fact that the full squares ( dots ) fit well with the empty circles ( squares ) . ] as the disorder strength @xmath11 increases but keeps in the range @xmath109 , the linear conductance keeps invariant .
but as @xmath110 , the chain is driven into the anderson localized phase , in which all the quasi - particle states are localized and then the chain is insulating . as a result
, the linear conductance abruptly drops to zero as @xmath11 crosses the critical point @xmath2 , as shown in the top panel of fig .
[ fig : transition ] .
we observe the transition from the superconducting phase to the insulating phase not only in the linear conductance but also in the energy spectra .
in the bottom panel of fig . [
fig : transition ] , we see that the superconducting gap gradually decreases as @xmath11 increasing .
the gap totally closes at the critical point , indicating the phase transition , and then reopens again .
the linear conductance keeps @xmath0 before the gap closes .
but beyond the critical point , the zero - energy modes ( the majorana bound states ) diminish , and the localized states open an energy gap @xcite .
the diminish of the zero - energy modes leads to a zero linear conductance . in fig .
[ fig : transition ] , we see that the drop of the linear conductance happens in the range @xmath111 $ ] , but not exactly at the critical point @xmath95 as expected by the phase transition theory . we ascribe this smooth transition to two reasons .
first , the kitaev chain in our calculation is finite with a length of @xmath112 , but the nonanalytic behavior happens only in the thermodynamic limit .
second , in our calculation we take the conductance at small voltage bias ( @xmath94 in fig . [
fig : transition ] ) to be approximately the linear conductance . around the critical point ,
the linear response regime is very small ( we do observe the shrink of the linear response regime in fig . [
fig : ivcurve ] as the disorder strength is closer to the critical point ) , so that the voltage bias that we choose is beyond the linear response regime .
we compare the current - voltage curves of the normal metal phase , the topological superconducting phase and the anderson localized phase in fig .
[ fig : ivcurve ] . beyond the linear response regime
, the @xmath1 curves also show abundant features that distinguish different phases . in the normal metal phase ,
the current increases linearly with the voltage bias in a wide range , and then turns into a plateau at the point @xmath113 , which is just the width of the conduction band of the chain .
the plateau indicates the limit of the current that the chain can carry . the linear response regime of the topological superconducting phase ( the curve titled @xmath104 in fig . [ fig : ivcurve ] ) is thinner than that of the normal metal phase . in the topological superconducting phase , as the voltage bias increasing ,
the current quickly runs into the first plateau associated with the superconducting gap . within the superconducting gap there
lie only a pair of majorana bound states exactly at zero energy , then the current will not increase any more as the voltage bias is beyond the width of the zero - mode levels since no new levels enter into the transmission window .
around @xmath113 , the current increases again and runs into the second plateau coinciding with that in the normal metal phase , due to the participation of the bulk states in the electron transport ( see the energy spectrum at @xmath83 and @xmath114 in fig .
[ fig : transition ] ) .
as the disorder strength increasing to @xmath81 , the linear conductance ( the slope of the @xmath1 curve at the origin ) keeps invariant , but the current at finite voltage is suppressed .
the two - plateau structure is also clear in the @xmath1 curve , but their heights are significantly smaller .
especially , the reduction of the current at the first plateau reflects the shrink of the linear response regime and then the width of the zero - mode levels .
the disorder narrows the linear response regime , indicating that purifying the sample can improve the temperature for observing the quantized conductance , which is comparable with the width of the zero - mode levels . in the presence of disorder ,
the left and right currents deviate from each other , even their slopes at the origin keep the same .
the reason of @xmath115 at finite voltage is that the p - wave pairing interaction breaks the conservation of the electron number ( @xmath85 is guaranteed at @xmath116 ) . at the same time
, the incommensurate potentials break the left - right symmetry of the kitaev chain . if rotating the chain by @xmath108 and then connecting it to the leads again , we see that the left and right currents exchange with each other ( see fig . [
fig : ivcurve ] ) . in the anderson localized phase
, the chain is insulating with all the quasi - particle eigenstates localized . at @xmath116 and @xmath81 , we see that the current is approximately zero in the whole range of voltage bias , reflecting the lack of a scattering state which can shuttle the electrons from the left lead to the right lead .
the conductance @xmath93 as a function of the gate voltage at @xmath104 and @xmath117 ( empty circles ) and at @xmath105 and @xmath94 ( solid circles ) .
the lines are the guides for the eyes . in both cases ,
the sharp drop of the conductance happens at the critical point dividing the topological superconducting phase and the topologically trivial superconducting phase .
[ middle panel ] energy spectra of the kitaev chain at @xmath104 under the open boundary condition .
[ bottom panel ] energy spectra of the kitaev chain at @xmath105 under the open boundary condition . ]
an important feature of the majorana state in the topological superconducting systems is that its energy is fixed to zero .
in contrast , the energy of the quasi - particle states in the topologically trivial phase can be adjusted by the gate voltage applied to the system .
for example , in the transport through a quantum dot in the coulomb blockade regime , the linear conductance depends on the gate voltage and shows a series of lorentzian peaks .
however , in our model , the energy of the majorana state is fixed inside the transmission window @xmath99 $ ] , then the resonant tunneling through it should not be affected by the gate voltage to the chain . this feature distinguishes the transport through the topological superconducting phase from that through the topologically trivial phase without zero modes . to verify this feature of the majorana state , we apply a non - zero gate voltage @xmath10 , and find that the linear conductance does keep the quantized value @xmath0 disregarding the gate voltage as it is in a certain range .
but the linear conductance abruptly drops to zero as @xmath10 is outside this range . in the top panel of fig .
[ fig : gatevoltage ] , we see that the conductance plateau is in the range @xmath118 $ ] at @xmath114 , while in the range @xmath119 $ ] at @xmath81 , as @xmath83 .
the plateau structure of the conductance function in fact reflects the quantum phase transition from the topological superconducting phase to the topologically trivial superconducting phase . in the middle and bottom panels of fig .
[ fig : gatevoltage ] , we show the energy spectra of the kitaev chain varying with @xmath10 at @xmath114 and @xmath81 respectively .
the common feature of the energy spectra is that there exists the zero modes accompanied by a finite energy gap in a range of @xmath10 . in this range ,
the chain is a topological superconductor with a pair of majorana bound states supporting the resonant tunneling .
the energy gap closes at the boundary of the topological superconducting phase , and then reopens as @xmath10 is outside the topological superconducting phase , accompanied by the diminish of the zero modes which results in a zero linear conductance . in fig .
[ fig : gatevoltage ] , we see that the points where the energy gap closes are @xmath120 for @xmath114 and @xmath121 for @xmath81 , coinciding well with the critical points where the conductance changes .
curves in the topological superconducting phase ( labeled by @xmath122 ) , in the topologically trivial superconducting phase ( labeled by @xmath123 ) , and in the anderson localized phase ( labeled by @xmath124 ) . ]
and we signal the topologically trivial superconducting phase also by its @xmath1 curve , which exhibits an exponential shape ( see fig . [
fig : gateivcurve ] ) .
the exponential @xmath1 curve always indicates a gap in the system , which is in fact the superconducting gap .
the shape of the @xmath1 curve in the trivial superconducting phase is quite different from that in the anderson localized phase , even the linear conductances in the two phases are both zero . in the superconducting phase , the chain can carry a strong current as the voltage bias overcomes the gap .
but in the anderson localized phase , the current keeps zero even at a large bias , because all the states are localized and can not carry the current .
in this paper , we employ the numerical operator method to study the transport through the kitaev chain with incommensurate potentials located between two normal leads . we find a quantized linear conductance of @xmath0 as the chain is in the topological superconducting phase , independent to the disorder strength or the gate voltage applied to the chain . by studying the energy spectra of the chain , we find that the quantized conductance results from the resonant tunneling via the majorana bound states lying exactly at the center of the transmission window and protected by an energy gap . while increasing the disorder strength or changing the gate voltage , at some critical point , the linear conductance abruptly drops to zero , corresponding to the transition of the chain into the anderson localized phase or the topologically trivial superconducting phase respectively . at the transition point , the energy gap closes , and then reopens with the zero modes diminishing .
the different phases can be distinguished by their @xmath1 curves .
the @xmath1 curve of the normal metal phase has a single plateau , while that of the topological superconducting phase has two , corresponding to the superconducting gap and the band edge respectively .
the linear response regime of the topological superconducting phase is small compared to that of the normal metal phase , and is further shallowed by the disorder .
the @xmath1 curves of the anderson localized phase and the topologically trivial phase both have a zero slope at the origin .
however , the latter exhibits an exponential shape signaling the superconducting gap , while the former is approximately zero in the whole range of voltage because all the quasi - particle states are localized .
p. wang has been supported by the nsf of china under grant no.11304280 .
s. chen has been supported by the nsf of china under grants no.11174360 and no.11121063 . and gao x. was supported by the nsf of china under grants nos .
11374266 and 11174253 and by the zhejiang provincial natural science foundation under grant no . r6110175 .
to calculate the ground - state properties of the kitaev chain , we re - express the hamiltonian ( [ hamiltoniansc ] ) in a matrix form : @xmath125 where the bogoliubov - de gennes matrix is written in blocks as @xmath126 with @xmath127 and @xmath128 .
we diagonalize the bogoliubov - de gennes matrix , and re - express the hamiltonian into @xmath129 , where @xmath17 and @xmath130 denotes the quasi - particle energy in the negative energy branch .
the energy spectrum @xmath131 and the coefficients @xmath19 and @xmath20 satisfy the eigen equation @xmath132 and are determined by a numerical diagonalization routine . in the ground state ,
the negative single - particle levels are all occupied while the positive ones are empty , so that @xmath133 as @xmath134 .
we then obtain the four types of correlation functions by using the inverse bogoliubov - de gennes transformation @xmath135 .
the correlation functions are expressed as @xmath25
, @xmath83 and @xmath136 . the left and right lead currents are the same in this case . ] to demonstrate the power of our method , we compare the result by the numerical operator method with that by the exact diagonalization method . in exact diagonalization , we take the length of each normal lead to be @xmath137 .
the currents by two different methods are plotted in fig .
[ fig : comparison ] .
it is clear that the current by numerical operator method coincides very well with that by exact diagonalization in short time period .
however , the current by exact diagonalization starts to oscillate as time increasing , the so - called finite size problem . while the _ i - t _ curve by numerical operator method still keeps smooth .
the numerical operator method conquers the finite size problem by setting two infinite leads at very beginning , and is especially appropriate for studying the currents in the presence of disorder when the long relaxation time of currents ( see fig .
[ fig : evolution ] ) prevents exact diagonalization from reaching the stationary currents .
h. haug and a .-
jauho , _ quantum kinetics in transport and optics of semiconductors _
( springer , berlin , 1996 ) .
w. g. van der wiel , s. de franceschi , t. fujisawa , j. m. elzerman , s. tarucha , and l. p. kouwenhoven , science * 289 * , 2105 ( 2000 ) .
k. t. law , p. a. lee , and t. k. ng , phys .
* 103 * , 237001 ( 2009 ) . c. w. j. beenakker , annu .
con . mat .
phys . * 4 * , 113 ( 2013 ) . c. j. bolech and e. demler , phys .
lett . * 98 * , 237002 ( 2007 ) .
k. flensberg , phys .
b * 82 * , 180516(r ) ( 2010 ) .
d. i. pikulin , j. p. dahlhaus , m. wimmer , h. schomerus , and c. w. j. beenakker , new j. phys . * 14 * , 125011 ( 2012 ) .
j. liu , a. c. potter , k. t. law , and p. a. lee , phys .
* 109 * , 267002 ( 2012 ) .
b. m. fregoso , a. m. lobos , and s. das sarma , arxiv:1307.3505 .
y. tanaka and s. kashiwaya , phys .
b * 70 * , 012507 ( 2004 ) .
g. moore and n. read , nucl .
b * 360 * , 362 ( 1991 ) .
t. m. rice and m. sigrist , j. phys .
c * 7 * , l643 ( 1995 ) .
j. nilsson , a. r. akhmerov , and c. w. j. beenakker , phys .
* 101 * , 120403 ( 2008 ) .
j. d. sau , r. m. lutchyn , s. tewari , and s. das sarma , phys .
* 104 * , 040502 ( 2010 ) .
l. fu and c. l. kane , phys .
lett . * 100 * , 096407 ( 2008 ) .
y. oreg , g. refael , and f. von oppen , phys .
. lett . * 105 * , 177002 ( 2010 ) .
p. a. ioselevich and m. v. feigelman , phys .
lett . * 106 * , 077003 ( 2011 ) .
f. s nogueira and i. eremin , j. phys . : condens .
matter * 24 * , 325701 ( 2012 ) . c. w. j. beenakker , d. i. pikulin , t. hyart , h. schomerus , and j. p. dahlhaus , phys .
. lett . * 110 * , 017003 ( 2013 ) .
a. y. kitaev , phys .
usp . * 44 * , 131 ( 2001 ) [ or arxiv : cond - mat/0010440 ( 2000 ) ] . c. nayak , s. h. simon , a. stern , m. freedman , and s. das sarma , rev .
mod . phys . * 80 * , 1083 ( 2008 ) .
v. mourik , k. zuo , s. m. frolov , s. r. plissard , e. p. a. m. bakkers , and l. p. kouwenhoven , science * 336 * , 1003 ( 2012 ) . m. t. deng , c. l. yu , g. y. huang , m. larsson , p. caroff , and h. q. xu , nano lett . * 12 * , 6414 ( 2012 ) .
a. das , y. ronen , y. most , y. oreg , m. heiblum , and h. shtrikman , nature physics * 8 * , 887 ( 2012 ) .
e. j. h. lee , x. jiang , m. houzet , r. aguado , c. m. lieber , and s. de franceschi , arxiv:1302.2611 [ cond-mat.mes-hall ] . y. asano and y. tanaka , phys . rev .
b * 87 * , 104513 ( 2013 ) .
s. aubry and g. andr , ann .
. soc . * 3 * , 133 ( 1980 ) .
g. roati , c. d errico , l. fallani , m. fattori , c. fort , m. zaccanti , g. modugno , m. modugno , and m. inguscio , nature ( london ) * 453 * , 895 ( 2008 ) . m. tezuka and a. m. garcia - garcia , phys .
a * 82 * , 043613 ( 2010 ) .
m. tezuka and n. kawakami , phys .
b * 85 * , 140508(r ) ( 2012 ) .
w. degottardi , d. sen , and s. vishveshwara , phys .
* 110 * , 146404 ( 2013 ) .
x. cai , l .- j .
lang , s. chen , and y. wang , phys .
lett . * 110 * , 176403 ( 2013 ) .
l. jiang , et .
al . , phys . rev .
* 106 * , 220402 ( 2011 ) .
i. c. fulga , a. haim , a. r. akhmerov , and y. oreg , new j. phys .
* 15 * , 045020 ( 2013 ) .
s. gangadharaiah , l. trifunovic , and d. loss , phys .
* 108 * , 136803 ( 2012 ) .
p. wang , physica e * 47 * , 141 ( 2013 ) ; arxiv:1209.3881 .
the gap in the bottom panel of fig .
[ fig : transition ] is quite small and does not show itself clearly , but the lowest levels in the anderson localized phase are in fact finite .
see the details in ref . [ 25 ] .
|
we study the transport through the kitaev chain with incommensurate potentials coupled to two normal leads by the numerical operator method .
we find a quantized linear conductance of @xmath0 , which is independent to the disorder strength and the gate voltage in a wide range , signaling the majorana bound states . while the incommensurate disorder suppresses the current at finite voltage bias , and then narrows the linear response regime of the @xmath1 curve which exhibits two plateaus corresponding to the superconducting gap and the band edge respectively .
the linear conductance abruptly drops to zero as the disorder strength reaches the critical value @xmath2 with @xmath3 the p - wave pairing amplitude and @xmath4 the hopping between neighbor sites , corresponding to the transition from the topological superconducting phase to the anderson localized phase .
changing the gate voltage also causes an abrupt drop of the linear conductance by driving the chain into the topologically trivial superconducting phase , whose @xmath1 curve exhibits an exponential shape .
|
investigation of the ultra - short high - intensity laser pulses propagation in air has been a hot topic in recent years due to its physical interest as well as its potential applications .
it is very important to predict well how the electromagnetic field of the pulse evolves as it propagates @xcite .
although some analytical solutions with some approximations can be found @xcite , in most applications the analytic approximations can not describe accurately the evolution of the pulses and we have to resort to numerical methods .
split - step fourier methods @xcite with the crank - nicholson scheme ( fcn ) in the transverse direction is often employed to numerical calculate the propagation of the ultra - short laser pulses via solving the nonlinear schr@xmath1dinger equation ( nlse)@xcite .
ultra - short high intensity laser pulse can convey high intensity over extended distances , and some applications need kilometer - range calculation .
the filamentation induced by the ultra - short high - intensity pulses have been observed over several kilometers @xcite .
the alternative signs in the coefficients of the high - order kerr effects allow the pulses to propagate without much energy loss in a long distance @xcite . however , the simulation for a long - distance propagation of the pulse is very time - consuming .
for example , it may need several weeks to calculate a kilometer - range propagation .
parallel split - step fourier methods @xcite have been developed to solve 1d+1 nlse .
however , the 1d+1 nlse can not be used to simulate ultra - short high intensity laser pulse propagation in air , since it is unable to describe the transverse variations of the laser pulse . in this paper
we develop a parallel fcn method to solve a 2d+1 nlse , which can be employed to simulate ultra - short high intensity laser pulse propagation in air .
the paper is organized as follows . in section 2
we briefly introduce the nlse which describes the ultra - short high - intensity laser pulses propagation in air . in section 3
the serial 2d+1 fcn method for solving the nlse equation is reviewed .
section 4 presents the parallel fcn algorithm for the 2d+1 nlse .
performance tests and a simulation for a long - distance pulse s propagation are given in section 5 .
conclusion is detailed in section 6 .
the wave equation for the laser electric field @xmath2 is given by @xcite @xmath3 where @xmath4 , and @xmath5 are the linear and nonlinear source of the electric field .
@xmath6 , @xmath4 , and @xmath5 can be written as @xcite @xmath7 where @xmath8 denotes the complex conjugate of the first term in the right hand side of equations .
@xmath9 is the carrier wave number , @xmath10 is the angular frequency of the pulse , @xmath11 denotes the unit vector in the direction of polarization .
@xmath12 , @xmath13 and @xmath14 are the complex amplitudes of @xmath6 , @xmath4 and @xmath5 respectively . for the ultra - short and high - intensity laser pulse
, @xmath15 can be written as @xmath16 where @xmath17 denotes the nonlinear contribution from bound electrons , i.e. , kerr effect , @xmath18 here @xmath19 is kerr nonlinear refractive index .
the nonlinear index defines a critical nonlinear self - focusing power @xmath20 for a gaussian input pulse . the plasma source term @xmath21 is given by @xmath22 where @xmath23 denotes the plasma frequency , @xmath24 is the plasma density generated by ionization , @xmath25 is the mass of electron and @xmath26 is the vacuum permittivity .
the term @xmath27 describes the depletion of laser energy due to ionization @xmath28 where @xmath29 is the coefficient of multiphoton ionization for the number of photons @xmath30 . when the wavelength is 800 nm , @xmath30 = 10 and @xmath31 + @xmath32 .
@xcite substituting eqs .
( 2)- into eq . and transforming the independent variables from @xmath33 to @xmath34 via @xmath35 with @xmath36 the linear group velocity of the pulse , and applying the slowly varying envelope approximation @xmath37 , we can obtain the nonlinear schr@xmath38dinger equation describing the propagation of ultra - short and high - intensity laser pulse as follow @xmath39 where @xmath40 for cylindrically symmetric beams or @xmath41 otherwise . in this paper
the initial input pulse is chosen to have cylindrical symmetry .
@xmath42 is the second order dispersion coefficient .
the rate equation for electron density @xmath24 can be written as @xmath43 where @xmath44 is the density of the neutral atoms .
the core contribution of this work is to built a 2d+1 ( time ) parallel fcn solver for eq . and eq ..
we restrict our attention to the axis - symmetric problems thus @xmath45 .
uniform discrete lattice approximations is employed with lattice spacings @xmath46 , @xmath47 , and @xmath48 .
@xmath49 is discretized into @xmath50(@xmath51+@xmath52 ) , in which @xmath53 , and @xmath54 . in fourier domain
, @xmath55 can be discretized into @xmath56 .
let @xmath57 denote @xmath58 and @xmath59 denote @xmath56 .
let @xmath60 represent the vector @xmath61 , @xmath62 represent the vector @xmath63 , and @xmath64 represent the matrix @xmath65 .
the nls equation can be written as @xmath66 where @xmath67 and @xmath68 are the linear and nonlinear operators respectively , @xmath69 in the step @xmath70 , we first calculate the nonlinear part for a half step ( @xmath71 ) , and then calculate the linear part for a full step @xmath72 , and finally calculate the nonlinear part for another half step ( @xmath71 ) , i.e. , @xmath73 in the below we describe the algorithm in detail . the linear part is solved in the frequency domain .
firstly , @xmath74 in the time domain are transformed to @xmath76 in the frequency domain . here
@xmath77 denote fourier transform .
it follows eq . that the linear - part effect satisfies @xmath78
secondly , we discretize eq . using crank - nicholson scheme which is an implicit finite - difference method and
is given by @xmath79 substituting eqs .
- into eq . , and making use of the boundary conditions @xmath80 we can obtain a matrix equation @xmath81 where @xmath82~,\\ \it{m_{def}^p } \equiv \left [ \begin{array}{cccccc } \it{e_{0 } } & \it{f_{0 } } & & \\
\it{d_{1 } } & \it{e_{1 } } & \it{f_{1 } } & \\
% & b_3 & a_3 & c_3\\ & \it{\ddots } & \it{\ddots } & \it{\ddots}\\ & & \it{d_{n_r-2}}&\it{e_{n_r-2}}&\it{f_{n_r-2}}\\ & & & \it{d_{n_r-1 } } & \it{e_{n_r-1 } } \end{array } \right]~,\end{aligned}\ ] ] with @xmath83}~,\nonumber\\ \label{matrixindexanrp } & & \it{a_{n_r-1}}=\it{0},~\nonumber\\ \label{matrixindexb1p } & & \it{b_{0}=1+\frac{i\delta z}{k_0\delta r^2}-\frac{ik^{''}\omega_p^2\delta z}{4}}~,\nonumber\\ \label{matrixindexbjp } & & \it{b_{j}=1+\frac{i\delta z}{2k_0\delta r^2}-\frac{ik^{''}\omega_p^2\delta z}{4}}~,~~\it{j\in[1,n_r-2]}~,\nonumber\\ \label{matrixindexbnrp } & & \it{b_{n_r-1}=1}~,\nonumber\\ \label{matrixindexc1p } & & \it{c_{0}=-\frac{i\delta z}{k_0\delta r^2}}~,\nonumber\\ \label{matrixindexcjp } & & \it{c_{j}=-\frac{i\delta z}{4k_0\delta r^2}\left(1-\frac{1}{2j}\right)}~,~~\it{j\in[1,n_r-2]}~,\nonumber\\ \label{matrixindexdjp } & & \it{d_{j}=\frac{i\delta z}{4k_0\delta r^2}\left(1-\frac{1}{2j}\right)}~,~~\it{j\in[1,n_r-2]}~,\nonumber\\ \label{matrixindexdnrp } & & \it{d_{n_r-1}=0}~,\nonumber\\ \label{matrixindexe1p } & & \it{e_{0}=1-\frac{i\delta z}{k_0\delta r^2}+\frac{ik^{''}\omega_p^2\delta z}{4}}~,\nonumber\\ \label{matrixindexejp } & & \it{e_{j}=1-\frac{i\delta z}{2k_0\delta r^2}+\frac{ik^{''}\omega_p^2\delta z}{4}}~,~~j\in[1,n_r-2]~,\nonumber\\ \label{matrixindexenrp } & & \it{e_{n_r-1}=0}~,\nonumber\end{aligned}\ ] ] @xmath84}~.\nonumber\end{aligned}\ ] ] finally , @xmath85 can be obtained from @xmath86 via inverse fourier transformation @xmath87 the pseudo - codes for the linear - part effects of the step @xmath70 are listed as follows : * first loop:@xmath88 @xmath89@xmath90 * end first loop * second loop:@xmath91 @xmath92 * end second loop * third loop:@xmath93 @xmath89@xmath94 * end third loop it is worth pointing out that the second loop involves a triangular matrix equation and thus can be solved efficiently via a chasing method .
the calculations of the nonlinear - part effects are divided into two stages , which can be calculated by @xmath95 and @xmath96 the calculations of eqs . and
involves the plasma density , which can be obtained via solving eq . with a fourth - order runge - kutta method .
the pseudo - codes for the nonlinear - part effects in the first half step are listed as follows : * outer loop:@xmath97 inner loop:@xmath91 @xmath89calculate electron density @xmath89calculate @xmath98 @xmath89@xmath99 end inner loop * end outer loop the pseudo - codes for the another half step are same as that for the first half step , and we do not repeat here . in summary
, the serial 2d+1 fcn method can be carried out by the following steps : 1 .
input the initial data , e.g. , the initial input pulse , the index of the nls equation , the ranges of time and space , and the grid steps @xmath100 .
2 . calculate the triangular matrix @xmath101 and @xmath102 for the discretized frequency @xmath103 in fourier domain .
3 . loop:@xmath104 calculate the first half step of the nonlinear part .
calculate the linear part , from @xmath105 to @xmath85 .
calculate the second half step of the nonlinear part .
suppose we have @xmath106 threads to carry out the fcn method in the simulations
. set the thread s i d to @xmath107 .
let @xmath108 and @xmath109 denote the begin grid number and the end grid number in the radial domain for the @xmath110 thread .
let @xmath111 and @xmath112 denote the begin grid number and the end grid number in the temporal domain for the @xmath113 thread . in order to achieve the optimal parallel efficiency ,
the four arrays @xmath114 and @xmath115 are set as follows @xmath116 where @xmath117 denotes the rounding - up ( ceiling ) operation , @xmath118 denotes the rounding - down ( floor ) operation , and @xmath119 denotes the modulo operation . in the step @xmath70 , the calculations for the discrete ( reverse ) fourier transforms and the nonlinear part are decomposed in the radial domain ( fig .
[ fig1 ] ( a ) ) , and the calculations for solving the linear equations are decomposed in the temporal domain ( fig .
[ fig1 ] ( b ) ) .
the pseudo - codes for the linear - part effects of the step @xmath70 in the @xmath113 thread are listed as follows : * first loop:@xmath120 @xmath89@xmath90 * end first loop * waiting until all other p-1 threads finish the first loop * second loop:@xmath121 @xmath92 * end second loop * waiting until all other p-1 threads finish the second loop * third loop:@xmath120 @xmath89@xmath94 * end third loop the pseudo - codes for the nonlinear - part effects of the first - half step @xmath70 in the @xmath113 thread are listed as follows : * outer loop:@xmath120 inner loop:@xmath91 @xmath89calculate electron density @xmath89calculate @xmath98 @xmath89@xmath99 end inner loop * end outer loop the pseudo - codes for the another half step are same as that for the first half step , and we do not repeat here .
the above parallel algorithm for the fcn method is built directly following the serial one . in order
to better suite parallel programming , we re - organize the parallel algorithm into three parts basing on the data decomposition ( see fig .
[ fig2 ] ) .
the pseudo - codes for the the @xmath122 step in the @xmath113 thread are listed as follows , * outer loop:@xmath120 inner loop:@xmath91 @xmath89calculate electron density @xmath89calculate @xmath98 @xmath89@xmath99 end inner loop @xmath123 * end outer loop * waiting until all other p-1 threads finish the corresponding loop * loop:@xmath121 @xmath92 * end loop * waiting until all other @xmath124 threads finish the corresponding loop * outer loop:@xmath120
@xmath125 inner loop:@xmath91 @xmath89calculate electron density @xmath89calculate @xmath98 @xmath89@xmath126 end inner loop * end outer loop suppose a global integer variable @xmath127 has the initial value @xmath128 , and a global mutex @xmath129 .
the waiting other @xmath124 threads can described as follows : * lock(@xmath129 ) @xmath130 * unlock(@xmath129 ) * loop if @xmath131==0 @xmath89break end if * end loop
the specifications of the computer which is used for the numerical experiments are as follows 1 .
software environment : operating system : centos 6.3 development platform : g++ , pthread 2 .
hardware environment : cpu : intel(r ) xeo(r ) e7 - 4807 @ 1.87 ghz cpu : cores : 24 the initial laser pulse we consider is assumed to be a gaussian beam @xcite @xmath132 where @xmath133 is the beam width , @xmath134 is the temporal half width , @xmath135 is the input peak intensity , and @xmath136 denotes the chirp of the incident pulse .
the performance of the parallel program is measured by speedup , which is defined as the ratio between sequential execution time and parallel execution time @xcite , @xmath137 a numerical example ( @xmath138 , and @xmath139 ) is tested with different thread numbers , and table 2 presents the comparison for the computational time with different thread numbers .
.computational time and speed - up ratio for different thread numbers .
[ cols="^,^,^",options="header " , ] for the accuracy , we have checked that all the simulation results of the parallel code with different threads are the same as that of the serial code .
we employ the parallel algorithm to simulate the propagation of the ultra - short laser pulse in air for 1.1 kilometers . in the simulation , @xmath140 w/@xmath141 , @xmath142 mm , @xmath143 fs , and @xmath144 .
the number of grids @xmath145 = 1350 , @xmath146 = 1024 , and @xmath147 = @xmath148 . fig .
[ fig4 ] presents the evolution of the on - axis intensity and the fluence profile of the beam .
it only takes the parallel program with 20 threads less than 3 days to do the simulation , in contrast , it would require about two months for a serial code to do the same work .
in this paper , a parallel 2d+1 fcn method is developed which has been tested on multi - core shared memory architectures . the simulation results shows that the speed - up ratio is more than 19.2 when the thread number is 20 .
the parallel fcn algorithm is of great importance in the simulations for the long - distance propagation of the ultra - short laser pulse , which is very useful to many applications such as lightning control , remote sensing , and so on .
this work was supported in part by the ph.d .
programs foundation of ministry of education of china grant no .
20110184110016 and the national basic research program of china ( 973 program ) grant no .
2013cb328904 , as well as the 2015 doctoral innovation funds of southwest jiaotong university .
a. chiron , b. lamouroux , r. lange , j .- f .
ripoche , m. franco , b. prade , g. bonnaud , g. riazuelo and a. mysyrowicz , `` numerical simulations of the nonlinear prpapgation of femtosecond optical pulses in gases '' , eur .
j. d 6 , 383 ( 1999 ) .
p. sprangle , j. r. pe@xmath149ano and b .hafizi , `` propagation of intense short laser pulses in the atmosphere '' , phys .
e 66 , 046481 ( 2002 ) .
b. yedierler , `` remote ionization by a short pulse laser beam propagating in the atmosphere '' , phys .
plasmas 15 , 073107 ( 2008 ) .
b. yedierler , `` nonlinear longigudinal compression of short laser pulses in the atmosphere '' , phys .
plasmas 16 , 053104 ( 2009 ) .
agrawal , nonlinear fibre optics , academic press , ny , 1989 .
g. m@xmath150chain , c. damico , y. andr@xmath151 , s. tzortzakis , m. franco , b. prade , a. mysyrowicz , a. couairon , e. salmon and r. sauerbrey , `` range of plasma filaments created in air by a multi - terawatt femtosecond laser '' , opt .
247 , 171 ( 2005 ) .
v. loriot , e. hertz , o. faucher and b. lavorel , `` measurement of high order kerr refractive index of major air components '' , opt .
express 17 13429 ( 2009 ) ; erratum in opt .
express 18 3011 ( 2010 ) .
p. b@xmath150jot , j. kasparian , s. henin , v. loriot , t. vieillard , e. hertz , o. faucher , b. lavorel and j .- p . wolf , `` higher - order kerr terms allow ionization - free filamentation in gases '' , phys . rev .
104 , 103903 ( 2010 ) .
x. xu and t. taha , `` parallel split - step fourier methods for nonlinear schrdinger - type equations '' , journal of mathematical modelling and algorithms 2 , 185 ( 2003 ) .
j. s@xmath152nchez - curto and p. chamorro - posada , on a faster parallel implementation of the split - step fourier method , parallel computing 34 , 539 ( 2008 ) .
t. r. taha and x. xu , `` paralel split - step fourier methods for the coupled nonlinear schrdinger type equations '' , the journal of supercomputing , 32 , 5 ( 2005 ) .
s. tzortzakis , l. berge , a. couairon , `` breakup and fusion of self - guided femtosecond light pulses in air '' , phys .
86 , 5470 ( 2001 ) .
michael j. quinn , parallel programming in c with mpi and openmp , the mcgraw - hill companies , inc . 2004 .
a. couairon , m. franco , g. m@xmath150chain , t. olivier , b. prade and a. mysyrowicz , `` femtosecond filamentation in air at low pressures : part i : theory and numerical simulations '' , opt .
259 , 265 ( 2006 ) .
|
a parallel 2d+1 split - step fourier method with crank - nicholson scheme running on multi - core shared memory architectures is developed to study the propagation of ultra - short high - intensity laser pulses in air .
the parallel method achieves a near linear speed - up with results for the efficiency of more than @xmath0 on a 24-core machine .
this method is of great potential application in studying the long - distance propagation of the ultra - short high intensity laser pulses .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore
|
the physical conditions inferred from the absorption features caused by high redshift damped ly@xmath0 absorption systems ( dlas ) are , in several aspects , similar to those in the interstellar medium of present - day galaxies .
it has therefore been suggested to identify dlas with the progenitors of such galaxies ( e.g. wolfe 1988 ) . at low redshift dlas
show a wide variety of morphologies ( le brun et al .
1997 ) . at high redshift , however , few have been detected in emission ( mller & warren 1995 ; warren & mller 1995 , djorgovski et al .
1996 , djorgovski 1997 , mller & warren 1998 ) and the nature of galaxies causing the absorption remains unclear for the majority of dlas at high redshift .
the main source of information on the nature of dlas comes from the associated metal absorption which probes the kinematics and the chemical enrichment history of the mainly neutral gas in these systems .
motivated by the characteristic asymmetric shape of the absorption profiles of low ionization species wolfe and collaborators ( e.g. wolfe 1986 ) have suggested that high - redshift dlas are large rapidly rotating discs akin to present - day spiral galaxies ( but see also ledoux et al .
1998 ) . in this picture
the generally low observed metallicity of high redshift dlas ( pettini 1994 , lu 1996 , pettini 1999 ) is taken as evidence that these are still chemically young ( lindner , fritze - von alvensleben & fricke 1999 ) .
llllllllll @xmath2 & @xmath3 & @xmath4 & @xmath5 & @xmath6 & @xmath4 & @xmath7&@xmath8&@xmath9&@xmath10 + + @xmath11 & @xmath12 & @xmath13 & @xmath14 & @xmath13 & @xmath15 & @xmath16&@xmath17&@xmath18&@xmath19 + + @xmath20 & @xmath21 & @xmath13 & @xmath14 & @xmath6 & @xmath15 & @xmath7&@xmath22&@xmath23&@xmath19 + + @xmath24 & @xmath3 & @xmath4 & @xmath5 & @xmath6 & @xmath15 & @xmath25&@xmath8&@xmath26&@xmath10 + the conjecture of dlas being large rotating discs is , however , at odds with the velocity width distribution , the size and the total cross section of rapidly rotating discs predicted by hierarchical cosmogonies ( haehnelt , steinmetz and rauch 1998 , hereafter hsr98 ) .
hydrodynamical simulations ( katz et al . 1996 ; haehnelt , steinmetz & rauch 1996 ; gardner et al .
1997ab ; rauch , haehnelt & steinmetz 1997 ) show that gas condensations leading to damped ly@xmath0 absorption do indeed occur in such scenarios .
however , the hierarchical cosmogonies predict that a significant fraction of the cross section for damped absorption is contributed by gas in halos with circular velocities as small as @xmath27 ( ma & bertschinger 1994 ; mo & miralda - escud 1994 ; klypin et al . 1995 , kauffmann 1996 , mo , mao & white 1999 ; gardener et al . 1999 ) , where we define the circular velocity as @xmath28 at a radius where the overdensity is equal to 200 .
in hsr98 we have shown that the shape of the absorption profiles is not unique but can be equally well produced by merging proto - galactic clumps ( cf .
nulsen , barcon & fabian 1998 ; mcdonald & miralda - escud 1999 , maller et al .
in hsr98 we further showed that the observed velocity width distribution of the absorption features of low ionization species can be reproduced in a standard cold dark matter ( scdm ) cosmology .
at the same time hierarchical cosmogonies such as scdm have been shown to reproduce the basic properties of the observed population of high - redshift galaxies ( adelberger et al .
1998 ; steidel et al . 1998 ; baugh et al . 1998 ; bagla 1998 ; jing & suto 1998 ; haehnelt natarajan & rees 1998 ; jing & suto 1998 ; contardo , steinmetz & fritze - von alvensleben 1998 ; steinmetz 1998 ; kauffmann et al . 1999 ; mo , mao & white 1999 ) . here
we establish a link between emission and absorption properties of high - redshift galaxies . in section 2
we briefly test if our previous findings depend on cosmology . in section 3
we predict the luminosity and impact parameter distribution of high - redshift galaxies responsible for dlas .
section 4 contains our conclusions .
in hierarchical cdm - like cosmogonies dlas arise naturally from the cool gas that accumulates at the center of dark matter ( dm ) halos ( e.g. katz et al .
1996 , gardner et al .
1997a / b ) .
absorption features of low ionization species like siii are generally believed to be good tracers of the motion of this gas in a gravitational potential well ( prochaska & wolfe 1997 , 1998 ) .
one major characteristic of these absorption features is their overall width which in hsr98 we have shown to be correlated with the circular velocity of the dm halo hosting the dlas . however , for a halo of given circular velocity a statistical distribution of velocity widths @xmath29 arises due to different orientations of the line - of - sight and different dynamical states of the gas of the dm halos . halos with small circular velocities are believed to lose most of their gas due to feedback effects . the velocity width distribution of dlas as probed by low ionization species can thus be written as @xmath30 in hsr98 we used simulated absorption profiles for a set of haloes with @xmath31 to investigate @xmath32 for a scdm model and found that the distribution does not depend explicitly on @xmath33 and depends little on redshift .
we therefore assume @xmath34 to be a function of @xmath35 only . in order to check if there is a dependence on cosmology we repeated our analysis for a @xmath36cdm model ( see table 1 for the assumed model parameters and hsr98 for details of the simulations ) .
figure 1 shows @xmath37 averaged over different halos and lines - of - sight .
there is little difference between the two cosmological models .
once the cross - section weighting for damped absorption is known the velocity width distribution is readily calculated from the distribution of circular velocities @xmath38 . it is currently difficult to infer this cross - section weighting reliably from numerical simulations .
gardner et al . ( 1997 ) e.g. found that the cross section of dm halos scales as a power - law of the circular velocity as @xmath39 . in their later work they found a much shallower scaling @xmath40 ( gardner et al .
this cross - section weighting is likely to be sensitive to the energy and momentum input due to supernovae which is difficult to include properly in these simulations .
here we take a different approach and use the observed velocity width distribution to determine the mean cross section weighting for damped absorption .
we also choose a power - law parameterization @xmath41 using the press - schechter formalism we calculate @xmath42 and obtain the velocity width distributions shown in figure 2 .
the observed distributions are well fit with @xmath43 ( see table 1 ) and @xmath44 .
note that there is little evolution with redshift in the observed as well as in the predicted distribution .
the values of beta are close to the old value but quite different to the new value of gardner et al .
( 1997,1999 ) .
recently , it has become possible to determine the luminosity function for the population of star - forming galaxies detected at high redshift ( steidel et al .
1996 , steidel et al .
1999 ) . the luminosity function at z=3 can be reproduced if a simple linear scaling of the luminosity in the @xmath45 band ( steidel et al . 1996 ) with mass of the dm halo is assumed , @xmath46\ ] ] ( haehnelt , natarayan & rees 1998 ) .
the required values of @xmath47 are also shown in table 1 .
the same simple model can also reproduce the clustering strength of high - redshift galaxies and its decrease with decreasing limiting uv luminosity of the galaxy sample ( but see somerville , primack & faber ( 1999 ) for a somewhat different model ) .
note that the linear scaling of the uv luminosity with mass does not necessarily imply that the star formation rate scales linearly with the mass of the dm halo as the dust extinction of the uv luminosity is probably luminosity dependent ( steidel et al .
if we make the plausible assumption that high - redshift galaxies are hosted by the same population of dark matter halos that are causing dlas and assume that the linear scaling of the uv luminosity with mass can be extrapolated to smaller masses we can predict the luminosity distribution of dlas .
all we need to know is the scaling of the cross - section of dm halos for damped absorption with circular velocity which we already inferred in the last section from the velocity width distribution .
the result is shown in figure 3
. 80 to 90 percent of dlas are predicted to be fainter than @xmath48 ( the spectroscopic limit ) , rather independent of cosmology .
some of these may be faint ly@xmath0 emitters ( fynbo , thomsen & mller 1999 ) but most of them will probably be very difficult to identify reliably .
similarly we can also predict the impact parameter distribution ( fig .
the overall rate of incidence , @xmath49 ( storrie - lombardi et al .
1996 ) , determines the normalization of the cross - section weighting .
the corresponding values of @xmath50 are given in table 1 .
the _ predicted impact parameters range between 0.3 and 1 arcsec _
, rather independent of cosmology .
the parameter @xmath51 , i.e. the radius at which the overdensity is equal to @xmath52 is also listed in table 1 .
the typical radii of the region causing damped absorption are generally about 25 to 50 percent of @xmath53 of the corresponding dm halo .
this is about a factor ten larger than the expected scale length of a centrifugally supported disc if the angular momentum of the gas is due to tidal torquing during the collapse of the dm halo ( mo , mao & white 1998 ) .
note that with the assumed scaling of the cross section @xmath54 increases more steeply with circular velocity ( @xmath55 ) than @xmath53 ( @xmath56 ) .
in hierarchical cosmogonies dlas are mainly caused by merging protogalactic clumps hosted by collapsed dm halos . the observed velocity width of the absorption profiles of associated low ionization species is well reproduced for a a range of cdm variants which reproduce the present - day space density of galaxy clusters . by combining the cross - section weighting inferred from the observed velocity width distribution with the mass luminosity relation inferred from the luminosity function and clustering properties of high redshift galaxies it is possible to link absorption and emission properties of dlas . about 10 to 20% of dlas
are predicted to be brighter than @xmath57 .
expected impact parameter typically range between 0.3 and 1 arcsec but there is a pronounced tail of larger impact parameters .
the predictions seem consistent with the luminosities and impact parameters of the small number of dlas with spectroscopically identified emission ( see e.g. mller & warren 1998 for an overview ) . in our model
these would be drawn from the tail of the distribution at the bright end and at large impact parameter , respectively .
the rather faint flux levels and small impact parameter predicted for the majority of dlas explains why searches for the emission from dlas have been notoriously difficult .
the predicted impact parameters are nevertheless about a factor three to five larger than the typical radii of the luminous regions of lyman - break galaxies .
thus , dlas at high redshift should preferentially sample the outer regions of galaxies at the faint end of the luminosity function .
this is consistent with the low observed metallicities in dlas and the outer regions of galaxies . in our model dlas
should not be expected to show the high metallicities estimated for lyman break galaxies ( pettini et al .
1999 ) ; the latter are derived from emission spectra , which are dominated by the light from the more highly enriched central regions .
hierarchical cosmogonies predict lyman - break galaxies with @xmath58 to become part of bright galaxies in galaxy clusters at the present day .
the fainter galaxies responsible for dlas should , however , be the building blocks of more typical ( @xmath59 ) present - day galaxies consistent with the findings from chemical evolution models for dlas .
most high - redshift dlas are predicted to be too faint to obtain spectra even with 10 m telescopes unless they are strongly magnified by gravitational lensing .
the analysis of their absorption properties is currently the prime method for studying the progenitors of normal present - day galaxies .
adelberger , k. l. , steidel , c. s. , giavalisco , m. , dickinson , m. , pettini , m. , & kellog , m. 1998 , apj , 505 , 18 bagla j. s. , 1998 , mnras , 297 , 251 baugh , c. m. , cole s. , frenk , c. s. , & lacey c. 1998 , apj , 498,504 contardo , g. , steinmetz , m. , & fritze - von alvensleben , u. 1998 , apj , 507 , 497 djorgovski , s. g. , pahre , m. a. , bechtold , j. , & elston , r. 1996 , nature , 382 , 234 djorgovski , s. g. 1997 , in : `` structure and evolution of the intergalctic medium from qso absorption lines '' , paris 1997 , eds .
p. petitjean and s. charlot , edition frontieres , p. 303
ferguson , a. m. n. , gallagher , j. s. & wyse , r. f. g. 1998 , aj , 116 , 673 fynbo , j. u. , thomsen , b. , & mller , p. 1999
, aa , in press , astro - ph/9911146 gardner , j. p. , katz , n. , hernquist , l. , & weinberg , d.h .
1997a , apj , 484 , 31 gardner , j. p. , katz , n. , weinberg , d.h . , & hernquist l. 1997b , apj , 486 , 42 gardner , j. p. , katz , n. , hernquist , l. , & weinberg d.h . 1999 ,
apj , submitted , astro - ph/9911343 giavalisco , m. , steidel , c. s. , & macchetto , f.d .
1996 , apj , 470 , 189 governato , f. , baugh , c. m. , frenk , c. s. , cole , s.,lacey , c. g. , quinn , t. , & stadel , j. 1998 , nature , 392 , 359 haehnelt , m. g. , steinmetz , m. , & rauch , m. 1996 , apj , 465 , l95 haehnelt , m. g. , steinmetz , m. , & rauch , m. 1998 , apj , 495 , 647 ( hsr98 ) haehnelt , m. g. , natarajan , p. , & rees , m. j. 1998 , mnras , 300 , 817 jing , j. p. , & suto , y. 1998 , 494 , l5 katz , n. , weinberg , d.h . , hernquist , l. , & miralda - escud j. 1996 , apj , 457 , l51 kauffmann , g. a. m. , & charlot , s. , 1994 , apj , 430 , l97 kauffmann , g. a. m. 1996 , mnras , 281 , 475 kauffmann , g. a. m. , colberg , j.m . , diaferio , a. , & white s.d.m .
1999 , mnras , 307 , 529 klypin , a. , borgani , s. , holtzman , j. , primack , j. 1995 , apj , 44 , 1 le brun , v. , bergeron , j. , boisse , p. , & deharveng j.m .
1997 , a&a , 321 , 731 ledoux , c. , petitjean , p. , bergeron , j. , wampler , e. j. & srianand r. 1998 , a&a , 337 , 51l lindner , u. , fritze- von alvensleben , u. , & fricke k.j . 1999 ,
a&a , 341 , 709l lu , l. , sargent , w. l. w. , barlow , t. a. , churchill , c. w. , & vogt , s. 1996 , apjs , 107 , 475 ma , c .- p .
, & bertschinger , e. 1994 , apj , 434 , l5 maller , a. h. , somerville r. s. , prochaska j. x. , & primack j. r. 1999 in : `` after the dark ages '' , astro - ph/9812369 mcdonald , p. , & miralda - escud , j. 1999 , apj , 519 , 468 mo , h. j. , & miralda - escud , j. 1994 , apj , 430 , 25 mo , h. j. , mao , s. , & white , s. d. m. 1998 , mnras , 295 , 319 mo , h. j. , mao , s. , & white , s.d.m . , 1999 , mnras , 304 , 175 mller , p. , & warren , s. j. p. , 1998 mnras , 299 , 661 mller , p. , & warren , s. j. 1995 , in _ galaxies in the young universe _ , eds .
hippelein , meisenheimer & rser , p88 nulsen , p. e. j. , barcons , x. , & fabian , a.c .
1998 , mnras , 301 , 168 pettini , m. , smith , l. j. , hunstead , r. w. , & king d. l. 1994 , apj , 426 , 79 pettini , m. 1999 , in:``chemical evolution from zero to high redshift '' , garching 1998 , eds .
walsh j. , rosa m. , astro - ph/9902173 pettini , m. , steidel , c.s .
, adelberger , k.l . ,
dickinson m. , & giavalisco , m. 1999 , apj , in press , astro - ph/9908007 prochaska , j. x. , & wolfe , a. m. 1997 , apj , 487 , 73 prochaska , j. x. , & wolfe , a. m. 1998 , apj , 507 , 113 rauch , m. , haehnelt , m.g . , & steinmetz , m. 1997 , apj , 481 , 601 steinmetz , 1998 , in stsci symposium series 11 , `` the hubble deep field '' , eds . livio , m. , fall , s.m . , madau , p. , pp . 168 somerville , r.s . ,
primack , j.r . & faber , s.m . , 1999 , submitted , astro - ph/9806228 steidel , c. s. , giavalisco , m. , pettini , m. , dickinson , m. , & adelberger , k. l , 1996 , apj , 462 , l17 steidel , c. s. , adelberger , k. , dickinson , m. , giavalisco , m. , pettini , m. , & kellog m. 1998 , apj , 492 , 428 steidel , c. s. , adelberger , k. , giavalisco , m. , dickinson , m. , & pettini , m. 1999 , apj , 519 , 1 storrie - lombardi , l. j. , mcmahon , r.g .
, & irwin , m.j .
1996 , mnras , 283 , l79 warren , s. j. , & mller , p. 1995 , in _ new light on galaxy evolution _ , iau symposium 171 , heidelberg june 1995 , in press wolfe , a. m. 1988 , in _ qso absorption lines : probing the universe _
, baltimore 1987 , cambridge university press wolfe , a. m. , turnshek , d. a. , smith , h. e. , & cohen , r. d. 1986 , apj , 61 , 249
|
we combine predictions for several hierarchical cosmogonies with observational evidence on damped ly@xmath0 systems to establish a correspondence between the high redshift galaxy population and the properties of damped ly@xmath0 systems ( dlas ) .
we assume that high redshift galaxies and damped ly@xmath0 systems are hosted by the same dark matter halos and require consistency between the predicted halo space density , the rate of incidence and the velocity width distribution of damped ly@xmath0 systems , and the observed galaxy luminosity function at the bright end .
we arrive at the following results : ( 1 ) predicted impact parameters between the damped absorption system and the luminous parts of the absorbing galaxy are expected to be very small ( 0.3 - 1 arcsec ) for most galaxies ; ( 2 ) luminosities of galaxies causing damped absorption are generally fainter than @xmath1 and damped ly@xmath0 systems are predicted to sample preferentially the outer regions of galaxies at the faint end of the galaxy luminosity function at high redshift .
therefore , dlas should currently provide the best probe of the progenitors of normal present - day galaxies .
= 24.5 cm [email protected] , [email protected] , [email protected]
|
in general , we can specify the performance of a linear optical device by defining the mode conversion efficiency between a set of input modes and output modes @xcite .
these modes are specified by the user , and kept fixed during the optimization process .
the input modes @xmath10 are at frequencies @xmath11 , and can be represented by equivalent current density distributions @xmath12 . the generated electric fields @xmath13 should satisfy maxwell s equations in the frequency domain , @xmath14 where @xmath15 is the electric permittivity , and @xmath16 is the magnetic permeability of free space .
we can then specify @xmath17 output modes of interest for each input mode @xmath18 .
the output mode electric fields @xmath19 are given over output surfaces @xmath20 , where @xmath21 .
the amplitude of each output mode should be bounded between @xmath22 and @xmath23 , which can be expressed as @xmath24 for @xmath10 and @xmath21 .
we are thus interested in finding @xmath15 and @xmath13 which simultaneously satisfy ( [ eqn : fsg_physics_spec_1 ] ) and ( [ eqn : fsg_physics_spec_2 ] ) .
the wdm grating was designed by specifying the input mode to be a @xmath25 diameter vertically - incident gaussian beam , and the output modes to be the fundamental te mode of the output silicon slab waveguides , as shown in figure * [ fig:1_fsg_invdes]a*. the algorithm was directed to maximize power into the left waveguide and minimize power into the right waveguide at @xmath26 , and the converse at @xmath27 . silicon layer on @xmath28 of buried oxide , in the pattern produced by the optimization algorithm ( see figure * 1b * ) . *
( b ) * intermediate steps in the optimization process , where @xmath29 is the iteration number .
the optimization process proceeds in two stages . in the first stage ,
the permittivity @xmath15 is allowed to vary continuously . in the second stage
, the design is converted to a binary level - set representation and fine - tuned . to clearly illustrate the design process , the diagrams are not to scale in the vertical direction .
[ fig:1_fsg_invdes ] ] given the design specifications , our algorithm iteratively optimizes the structure using a close analogue of steepest - descent optimization to meet the constraints given in equation ( [ eqn : fsg_physics_spec_2 ] ) , as detailed in the supplementary information .
we use finite - difference frequency domain ( fdfd ) simulations to calculate the local gradient during each step @xcite . as illustrated in figure * [ fig:1_fsg_invdes]b * , the design procedure consists of two stages . in the first stage , the permittivity @xmath15 is allowed to smoothly vary within the design region . in the second stage , we convert the structure to a level - set representation @xcite and fine - tune the final structure
. we did not apply any design rules such as a minimum feature size , although such constraints could be incorporated into the design process @xcite .
the entire inverse design process for this grating took only @xmath30 minutes on a single intel core i7 processor .
the final device was fabricated in soi with a @xmath31 thick si device layer and a @xmath28 box ( buried oxide ) layer by fully etching the si device layer , as described in figure * [ fig:2_fsg_struct]a*. the dimensions and locations of the grating trenches are detailed in supplementary table * 1*. the broadband performance of the device was verified by using 2d finite - difference time domain ( fdtd ) simulations .
the phasor fields at @xmath32 and @xmath33 for a @xmath34 diameter gaussian beam , obtained using 2d - fdtd simulations , are plotted in figure * [ fig:2_fsg_struct]b*. the fabricated device is presented in figure * [ fig:2_fsg_struct]c*. the final fabricated structures also incorporate a waveguide and curved broadband output grating on either side of the wavelength - demultiplexing grating , as shown in figure * [ fig:3_fsg_expconfig]a*. scanning electron microscopy ( sem ) images of the fabricated structures are shown in figure * [ fig:3_fsg_expconfig]b*. the output gratings were strongly chirped to provide broadband performance . to minimize fabry - perot resonances due to back - reflections ,
the output gratings were slightly curved and placed far from the ends of the waveguides ( inset ) . in future , the output waveguides could be edge - coupled to optical fibers to obtain a well - characterized out - coupling efficiency @xcite .
the frequency - splitting grating was excited in the vertical direction by a focused gaussian beam with a diameter of @xmath35 , and a fraction of the coupled light was out - coupled by the output gratings . to ensure a clean gaussian input beam ,
the source was passed through a length of single - mode optical fiber .
spectroscopic data was measured using a broadband led source , and narrowband images were taken using tunable lasers .
the structure was both excited and measured through a single plan - apochromat microscope objective integrated into a custom confocal microscopy setup .
the collected light was either directly imaged using an ingaas near - infrared camera , or spatially filtered by a pinhole at a focal plane and analyzed using a grating spectrometer with an ingaas detector .
images of the device broadly illuminated with white light , and excited by a focused tunable laser are presented in figure * [ fig:3_fsg_expconfig]c - e*. at @xmath36 , light is only coupled to the left output grating , and at @xmath33 , light is only coupled to the right output grating , clearly demonstrating the basic functionality of the device . in figure * [ fig:5_fsg_expdata ] * , we present both the simulated and experimentally measured coupling efficiency spectra of the frequency - splitting grating .
the coupling efficiencies of the frequency - splitting grating computed using 2d fdtd are plotted in figure * [ fig:5_fsg_expdata]a*. at @xmath32 , the simulated coupling efficiencies into the left and right waveguides are @xmath37 and @xmath38 respectively , whereas at @xmath33 , the coupling efficiencies are @xmath39 and @xmath40 . the measured signal intensity , normalized to source brightness , from the two output gratings is plotted in figure * [ fig:5_fsg_expdata]b*. the output grating coupling efficiencies were not measured , but due to the symmetry of the experimental setup , measurements of the two output ports are directly comparable . the experimental data from @xmath41 is shaded to due to the presence of strong atmospheric water absorption lines @xcite .
fabry - perot fringing , arising from reflections between the frequency - splitting grating and output gratings , is visible in the measured spectra , with fringe spacing corresponding to the spacing between the gratings .
the measured signals broadly match the simulated coupling efficiencies of the frequency - splitting grating .
a*. * ( a ) * simulated coupling efficiency into the left and right waveguides , calculated using an finite - difference time domain ( fdtd ) simulation on the structure with optimized parameters .
a @xmath25 diameter gaussian beam was used as the input . *
( b ) * experimental measured intensities from the left and right output gratings .
the intensities are only normalized with respect to the source since the output grating efficiencies are not known .
around @xmath26 , light is predominantly coupled to the left grating , whereas around @xmath27 , light is predominantly coupled into the right grating .
measurements in the @xmath42 band ( shaded ) are corrupted by water absorption lines in the atmosphere . *
( c ) * simulated and experimentally measured splitting ratios for the coupler , defined as the the ratio of power into the left and right waveguides .
we have experimentally measured splitting ratios of @xmath43 at @xmath44 and @xmath45 at @xmath33 .
[ fig:5_fsg_expdata ] ] the splitting ratio , defined as the ratio of power emitted from the two output ports , is plotted in figure * [ fig:5_fsg_expdata]c*. the fabricated device has a measured splitting ratio of @xmath43 at @xmath44 and @xmath45 at @xmath33 , whereas the designed values at these wavelengths were @xmath46 and @xmath47 and respectively .
the discrepancy between the simulated and measured splitting ratio is likely due to fabrication imperfections . in conclusion ,
we have provided the first experimental demonstration of a nanophotonic device designed using our inverse design algorithm . in particular , we have implemented an efficient vertical - incidence wavelength - demultiplexing grating coupler , which can not be designed by hand or by using parameter sweeps . by vastly opening up the parameter space for nanophotonic devices ,
inverse - design has broad implications for the future design of novel and compact nanophotonic components with full three - dimensional freedom .
our inverse design algorithm uses a close analogue of steepest - descent optimization , and incorporates the maxwellfdfd finite - difference frequency - domain solver to calculate local gradients @xcite , as detailed in the supplementary information . after completing the design process ,
the broadband performance of the device was calculated using 2d finite - difference time - domain ( fdtd ) simulations .
the detailed grating dimensions are listed in supplementary table * 1*. the devices were fabricated on unibond@xmath48 smartcut@xmath48 silicon - on - insulator ( soi ) wafers obtained from soitec , with a nominal @xmath31 device layer and @xmath49 box layer .
a jeol jbx-6300fs electron beam lithography system was used to pattern @xmath50 of zep-52a electron beam resist spun on the samples .
the pattern was then transferred to the si device layer with a magnetically - enhanced reactive - ion etcher using a @xmath51 chemistry .
finally , the mask was stripped by sonicating in microposit remover 1165 .
this work has been supported by the afosr muri for complex and robust on - chip nanophotonics ( dr . gernot pomrenke ) , grant number fa9550 - 09 - 1 - 0704 .
a.y.p . also acknowledges support from the stanford graduate fellowship .
kgl acknowledges support from the swiss national science foundation .
a.y.p . designed and fabricated the structures , simulated their performance , and performed the experiments . j.l . developed the inverse - design algorithm used in this paper .
j.v . and t.b .
provided theoretical and experimental guidance , and j.v . supervised the entire project
. k.g.l . and j.p .
provided experimental support .
all authors contributed to discussions .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . &
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
& _ _ ( , ) . & . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ _ ( , ) . , & . _ _ * * , ( ) . & .
_ _ * * , ( ) .
( ) . https://github.com/wsshin/maxwellfdfd .
& _ _ ( , ) .
_ et al . _ . _
_ * * , ( ) .
maxwell s equations in the frequency domain can be written as @xmath52 where @xmath53 is the electric field , @xmath54 is the current density , @xmath55 is the frequency , @xmath15 is the electric permittivity , and @xmath16 is the magnetic permeability of free space .
we specify the device performance by defining the mode conversion efficiency between a set of input modes and output modes .
the input and output modes are specified by the user , and kept fixed during the optimization process .
the input modes @xmath10 are at frequencies @xmath11 , and can be represented by equivalent current density distributions @xmath12 .
we can then specify @xmath17 output modes of interest for each input mode @xmath18 .
the output mode electric fields @xmath19 are given over output surfaces @xmath20 , and the amplitude of each output mode should be bounded between @xmath22 and @xmath23 , where @xmath21 .
we are thus interested in finding @xmath15 and @xmath13 which simultaneously satisfy @xmath56 for @xmath10 and @xmath21 .
the permittivity @xmath15 is also subject to additional fabrication constraints .
this can be recast in the language of linear algebra by discretizing space and making the substitutions @xmath57 this leaves us with the problem @xmath58 for @xmath10 and @xmath21 . here , @xmath59 refers to the diagonal matrix whose diagonal entries are given by the vector @xmath60 . for convenience
, we further define @xmath61 which lets us rewrite equation ( [ eqn : fsg_maxwell_linalg ] ) as @xmath62 the final problem we wish to solve is then @xmath63 we have previously developed two inverse design methods for designing linear optical devices : one which we call * objective - first * , and another which is an analogue of the * steepest - descent * strategy @xcite . to design the wdm grating device demonstrated in this paper , we only used the steepest - descent based method , which is what we describe here .
the particular optimization problem we solve is @xmath64 here , we constrain the fields to satisfy maxwell s equations , parameterize the permittivity @xmath65 with @xmath66 , and construct a penalty function @xmath67 for violating our field constraints from equation ( [ eqn : fsg_prob_constr ] ) .
the penalty @xmath68 for each input mode is given by @xmath69 where @xmath70 is a relaxed indicator function , @xmath71 typically , we use @xmath72 and @xmath73 .
we ensure that maxwell s equations always satisfied , which implies that both the fields @xmath74 and the field - constraint penalty function @xmath75 are a function of the permittivity @xmath65 . on each iteration , we locally approximate our penalty function @xmath76 with a quadratic function @xmath77 and solve the subproblem @xmath78 using steepest descent optimization .
the structure parameter @xmath79 is sufficiently small that we can calculate the gradient @xmath80 using brute force .
the main computational cost of the algorithm lies in computing @xmath81 .
we choose a quadratic function @xmath81 of the form @xmath82 where @xmath83 is the value of @xmath65 from the previous iteration in the optimization process , and @xmath84 .
the analogy with the steepest descent strategy is clear if we consider the minimum of @xmath81 , @xmath85 which is simply the steepest descent step with size @xmath86 .
we now consider how to compute the gradient @xmath87 . taking the derivative of equation ( [ eqn : fsg_maxwell_linalg ] ) , the discretized maxwell s equations , with respect to @xmath65
, we obtain @xmath88 where we have used our definitions of @xmath89 and @xmath90 from ( [ eqn : fsg_defab ] ) .
the derivative of @xmath91 with respect to @xmath65 is then given by @xmath92 the gradient of the structure objective is thus @xmath93 where @xmath94 since @xmath89 and @xmath90 are large @xmath95 matrices , we have rearranged the expression in the final step to require only a single matrix solve rather than @xmath29 solves .
the partial derivatives @xmath96 are given by @xmath97 where @xmath98 the absolute value function @xmath99 for @xmath100 is not analytic so the complex derivative does not exist .
instead , we have used the wirtinger derivative @xcite of the absolute value function in ( [ eqn : fsg_didx_1 ] ) - ( [ eqn : fsg_didx_2 ] ) , which is @xmath101 where we have defined @xmath102 for @xmath103 .
the particular structure we designed consisted of two materials in the design area , with permittivities @xmath104 and @xmath105 .
we initially used a linear parameterization of @xmath106 , @xmath107 where @xmath108 , @xmath109 , and @xmath110 .
the portions of the structure which were kept fixed during the optimization process were described by @xmath111 . in the second step of the optimization , we converted to a level set representation of the structure using thresholding .
when constructing @xmath65 , we took care to apply anti - aliasing to the borders of the structure .
we must efficiently solve maxwell s equations in the frequency domain at the following points in our algorithm : 1 . evaluating the fields @xmath91 at the beginning of each iteration .
2 . solving the adjoint problem in equation ( [ eqn : fsg_dfi_dxi ] ) to compute the gradient @xmath87 of the field - constraint penalty .
this was done by using the maxwellfdfd package for matlab @xcite .
1 url # 1`#1`urlprefix[2]#2 [ 2][]#2
|
nanophotonics has emerged as a powerful tool for manipulating light on chips .
almost all of today s devices , however , have been designed using slow and ineffective brute - force search methods , leading in many cases to limited device performance . in this article , we provide a complete demonstration of our recently proposed inverse design technique , wherein the user specifies design constraints in the form of target fields rather than a dielectric constant profile , and in particular we use this method to demonstrate a new demultiplexing grating .
the novel grating , which has not been developed using conventional techniques , accepts a vertical - incident gaussian beam from a free - space and separates o - band @xmath0 and c - band @xmath1 light into separate waveguides .
this inverse design concept is simple and extendable to a broad class of highly compact devices including frequency filters , mode converters , and spatial mode multiplexers .
@xmath2 ginzton laboratory , stanford university , stanford , ca , 94305 @xmath3 [email protected]_ conventional integrated photonic devices @xcite include components such as waveguide directional couplers @xcite , multimode interference couplers @xcite , distributed bragg reflectors @xcite , micro - ring resonators @xcite , adiabatic tapers @xcite and grating couplers @xcite . in all of these cases ,
the design space spans a relatively small @xmath4 number of parameters such as structure widths , heights and periodicity that are tuned throughout the photonic device design stage . to design a device
, the photonic engineer specifies a dielectric profile , computes the electromagnetic field response using maxwell s equations , and compares the response to the device specifications . the process is then repeated , modifying the dielectric profile each iteration , until satisfactory performance is obtained .
this brute force approach suffers from a long device design cycle , and does not take full advantage of the available design space of fabricable devices . to this end , a wide variety of increasingly sophisticated approaches have been developed to search through this parameter space and optimize specific nanophotonic structures .
several methods , namely genetic algorithms @xcite and particle swarm optimization @xcite , ignore the underlying physics but have achieved considerable success in fine - tuning existing structures , and designing photonic crystal devices with selectively removed holes and posts .
these methods , however , are typically restricted to optimizing a relatively small number of geometric parameters , and scale poorly with additional degrees of freedom .
other methods exploit the underlying physics to quickly converge on local optima , typically by computing the local gradient of a performance metric and using steepest - descent optimization @xcite . owing to their much faster convergence
, they can be used to design more complex structures with arbitrary topologies @xcite .
these algorithms can search through the design space of complex , aperiodic structures beyond those we can come up with based on our intuition and experience .
such devices may be able to provide novel functionality , or higher performance and smaller footprints than traditional devices , due to the greatly expanded design space .
we have recently proposed an inverse design approach for linear optical components , where the user specified input is not the basic structure of the device , but rather a set of performance metrics @xcite .
these include the device area , the modes of input and output waveguides to use ( e.g. te , tm , and mode order ) , extinction ratios ( e.g. @xmath5 ) , and insertion loss ( e.g. @xmath6 ) . more rigorously , the user only specifies the coupling efficiencies between a set of input and output modes at various frequencies . any linear optical device can be specified in this fashion @xcite , including mode converters , spatial mode multiplexers , and wavelength demultiplexers .
two of the most important functions in integrated photonics , for both chip - to - chip and intra - chip optical interconnects , are wavelength - division multiplexing ( wdm ) and vertical - incidence coupling @xcite . a device combining these two functions in a single , compact layout could be particularly useful for coupling between silicon photonic layers in a stacked - die microprocessor , or for coupling on / off chip using optical fibers .
although a uniform grating coupler with a tilted incident beam will act as a wavelength - demultiplexing grating coupler for two wavelength bands @xcite , the symmetry imposed by a vertically incident beam implies that a uniform grating coupler can not split wavelengths .
indeed , an efficient vertically - incident wavelength - demultiplexing grating coupler can not be designed using current analytic methods , or by tuning a small number of parameters by hand . here
, we provide a full demonstration of the inverse design technique s power by experimentally demonstrating a vertical - incidence , wavelength - demultiplexing grating coupler fabricated in silicon - on - insulator ( soi ) .
the grating accepts a vertically incident beam from free space or an optical fiber , and splits o - band @xmath7 and c - band @xmath8 light into separate silicon photonic waveguides with high extinction ratios @xmath9 .
|
mid - infrared ( mid - ir ) spectra of active galactic nuclei ( agns ) are rich in emission features , such as polycyclic aromatic hydrocarbons ( pahs ) , molecular hydrogen and prominent forbidden emission lines @xcite .
another spectral feature commonly observed in agns is silicate at @xmath12 9.7@xmath0 m and 18@xmath0 m , both in emission and absorption .
the agn unified model proposes the existence of a dense concentration of absorbing material surrounding the central engine in a toroidal distribution , which blocks the broad line region ( blr ) emission from the line of sight in seyfert 2 ( sy2 ) galaxies @xcite .
however , it is not yet clear if the molecular gas and dust detected in mid - ir spectra is actually associated with the absorbing material required by the so - called torus in the unified model .
@xcite studied the distribution of the silicate feature strengths in a sample of 196 agns and ultraluminous infrared galaxies ( ulirgs ) .
they found that the spectra of quasars and seyfert 1 ( sy1 ) galaxies are characterized by silicate features in emission , with few sy1s presenting weak absorptions .
in contrast , sy2 spectra are dominated by weak silicate absorption .
these results suggest that the silicate emission ( or absorption ) is related to the line of sight , in the framework of the agn unified model .
in addition , most of the ulirgs show very deep silicate absorption @xcite . the silicate behavior in agns can be explained using nenkova s torus models @xcite , which consider an ensemble of individual clouds instead of the standard `` solid doughnut '' .
@xcite detected the silicate emission feature in the sy2 galaxy ngc2110 , and compared it with the @xcite clumpy torus models , demonstrating that the presence of silicate emission can also be explained by an edge - on clumpy distribution . from their best - fit model , they conclude that there would be a small number of clouds along the line of sight , with a distribution that does not extend far above the torus equatorial plane .
according to @xcite the presence of strong silicate absorption and the absence of pah features in some active galaxies , can be interpreted as extremely heavily obscuration in these sources .
an example of such an object is ngc3281 .
its spectrum shows a very deep silicate absorption at 9.7@xmath0 m , as well as the forbidden emission lines [ siii]18.7@xmath0 m , [ siv]10.5@xmath0 m , [ neii]12.8@xmath0 m , [ neiii]15.5@xmath0 m , [ nev]14.3@xmath0 m and 24.3@xmath0 m , [ nevi]7.6@xmath0 m and [ oiv]25.8@xmath0 m . no pah emission was detected in the spitzer spectrum @xcite .
in addition , ngc3281 is a luminous infrared galaxy ( logl@xmath13 10.73l@xmath14 , * ? ? ? * ) , classified as a sy2 .
its optical spectrum was studied by @xcite , finding that the kinematics of the kpc extended ionized gas can be described by rotation in a plane , with material outflowing ( @xmath15150 km s@xmath16 ) from the nucleus within a conical structure .
they also suggest that ngc3281 has dust asymmetrically distributed with respect to the ionized gas .
in addition , using advanced satellite for cosmology and astrophysics ( _ asca _ ) data , @xcite found that ngc3281 shows an extremely complex x - ray spectrum , with a prominent iron emission line and a large absorbing column density ( n@xmath17 [email protected] @xmath19 10@xmath20@xmath21 ) that when combined with the nuclear extinction inferred from the near - infrared color analysis ( a@xmath22 = 22@xmath1811mag ) , results in a n@xmath23/a@xmath22 ratio one order of magnitude higher than that of the galaxy @xcite .
this effect was attributed to an optically thick material along the line of sight of both the x - ray and the infrared emitting regions , which obscures the entire x - ray source but only a fraction of the much larger ir emitting region .
later , @xcite using the italian - dutch satellite _ bepposax _ , found a n@xmath23/a@xmath22 ratio about twice that derived by @xcite , adding ngc3281 to the list of agns presenting dust properties different from the galaxy .
the difference in the n@xmath23/a@xmath22 values derived by both studies , according to @xcite , is due to the limited _ asca _ band - pass .
@xcite also show that its x - ray spectrum ( e@xmath2410kev ) is reflection - dominated , with a relatively strong k@xmath10 iron emission line and a heavily absorbed nuclear continuum ( n@xmath25 10@xmath26@xmath21 ) .
hence , they classify ngc3281 as a compton - thick galaxy . the high values found for the absorption column density in ngc3281 also explain the extremely high ratio f@xmath27\,}}$]/f@xmath28 @xmath11 0.3 , when compared to the average ( @xmath11 0.02 ) value obtained for a sample of sy2 galaxies @xcite .
it follows from the above discussion that this galaxy has a very heavily obscured nucleus , which makes it a key object to study whether or not the observed dust absortion is associated with the dusty torus of the unified model .
thus , it may provide unambiguous observational evidence of this structure .
we present in this paper ground bases mid - ir high - angular resolution spectra of the galaxy ngc3281 .
the thermal - region camera spectrograph ( t - recs ; * ? ? ?
* ) attached to the gemini south telescope give a spatial resolution of 16 pc / pix , which is very adequate to study the dust distribution in the @xmath12 200pc central radius of this galaxy .
this paper is structured as follows : in section [ observation ] we describe the observation and data reduction processes .
results are discussed in section [ results ] .
summary and conclusion are presented in section [ conclusions ] .
ground - based mid - ir observations of ngc3281 were obtained with the t - recs in queue mode at gemini south , in 2009 april 06 ut , as part of program gs-2009a - q-34 .
conditions were clear , with water vapor column in the range 58 mm , and image quality of 033 in the n band , as measured from the acquisition images of the telluric standards observed right before and/or after the science frames .
all observations used a standard chop - nod technique to remove time - variable sky background , telescope thermal emission , and the effect of 1/_f _ noise from the array / electronics .
the slit was oriented along pa = 315 , with a chop throw of 15 , oriented along pa = 225 , perpendicular to the slit direction , in order to include only the signal of the guided beam position in the frame and avoid possible nod effects in the spatial direction .
the same slit position / nod orientation were used for the telluric standards .
the instrument configuration used the low - resolution ( r @xmath12 100 ) grating and the 036 slit , for a dispersion of 0.0223@xmath0m / pixel and a spectral resolution of 0.08@xmath0 m .
the pixel scale is 0.089 arcsec / pixel in the spatial direction , and the slit is 216 long .
spectral coverage was of 7.1@xmath0 m centered at 10.5@xmath0 m .
the total on - source integration time was 980 sec .
the data reduction was performed using the midir and gnirs sub - packages of gemini iraf package . to extract the final spectrum , we combined the chop- and nod - subtracted frames using the tasks tprepare and mistack in the midir package .
wavelength calibration was obtained from the sky lines in the raw frames . to remove the telluric absorption lines from the galaxy spectrum , we divided the extracted spectrum for each observing night by that of the telluric standard stars hd3438 or hd4786 @xcite , observed before and/or after the science target . finally , the spectrum was flux calibrated by interpolating a black - body function to the spectrum of the telluric standard . for this step we used the task mstelluric in the midir package .
[ fenda]a shows the slit superposed on the ngc3281 n - band acquisition image .
panel b of this figure illustrates the slit on the narrow [ oiii ] maps taken from @xcite .
panel c shows that the n - band galaxy luminosity profile is extended with respect to that of the n - band stellar profile of hr3438 .
we extracted seven one dimensional ( 1d ) spectra : one centered in the unresolved nucleus ( 4-pixel = 0.36 arcsec , which corresponds to @xmath1265 pc for a distance of 43 mpc , using h@xmath29 = 74 km s@xmath16 mpc@xmath16 ) , plus extractions centered at 130pc , 195pc and 260pc to the northwest ( nw ) and southeast ( se ) directions .
deep silicate absorption is observed at the nucleus and at 130pc in both nw and se directions of the nucleus ( fig .
[ spectra ] ) .
the extraction at 195pc nw shows a weak absorption . the remaining extraction does not present silicate absorption .
therefore , we can conclude that the dust is concentrated inside a radius of 200pc .
the ngc3281 t - recs spectra ( fig . [ spectra ] ) clearly show that a deep silicate absorption , as well as the [ siv]10.5@xmath0 m and [ neii]12.7@xmath0 m emission lines , are observed up to 130 pc from the nucleus .
in addition , pah emission bands are not detected in the spectra , in agreement with the spitzer spectrum @xcite . as discussed in the introduction , ngc3281 is one exception among sy 2 galaxies , as most ( @xmath1280% ) of the objects in this class present pah emission bands and no deep silicate absorption in their mid - ir spectra @xcite . in order to estimate the silicate absorption apparent strength ( @xmath30 ) ,
we use the definition of @xcite : @xmath31 where @xmath32 is the observed flux density and @xmath33 is the observed mid - line pseudo - continuum flux density . to determine the s@xmath34
, we avoid telluric bands and emission line regions , using the same methodology as @xcite .
the values derived from the four spatial positions where the silicate absorption is observed are shown in column 7 of tab .
[ lines ] . for the unresolved nucleus , we obtain s@xmath34 = @xmath35 .
similar values of s@xmath34 are observed in ulirgs , while much weaker ( average is -0.61 ) absorptions are seen in sy2s ( see fig.2 of * ? ? ?
the remaining three extraction have similar values to that of the nucleus .
when dealing with dusty sources , a more physical parameter is the line of sight depth ( @xmath36 ) , which can be estimated from s@xmath34 at 9.7@xmath0 m using the equation @xmath37 @xcite .
furthermore , we can derive the apparent optical extinction using a@xmath38 = @xmath3918.5@xmath182 mag @xcite .
the resulting values are shown in tab .
[ lines ] .
the dust extinction for the unresolved nucleus of ngc3281 is a@xmath38 = 24@xmath185mag , in agreement with the value ( a@xmath2 = 22@xmath1811 mag ) derived by combined ir data @xcite .
peruse that our uncertainties is much lower than these authors .
we note from the values listed in tab .
[ lines ] that the position labeled as 1 ( 130pc from the nucleus at nw direction ) presents a slightly higher extinction than the central one ( a@xmath2 = 28@xmath188 mag ) . to estimate in a more robust way the line of sight depth at 9.7 ( @xmath40 ) in ngc3281 we used the pahfit code @xcite , which assumes that the source spectrum is composed by continuum emission from dust and starlight , prominent emission lines , individual and blended pah emission bands , and that the light is attenuated by extinction due to silicate grains .
pahfit models the extinction using the dust opacity law of @xcite , where the infrared extinction is considered as a power law plus silicate features peaking at 9.7@xmath0 m . for details see @xcite .
we used this approach to determine the contribution of the distinct components to the spectral energy distribution ( sed ) of ngc 3281 as a function of the distance to the galaxy centre .
as the spectra do not show pah emission , we have not included this component in the fitting .
the other input parameters are the same as those used by @xcite , which are appropriate for agns .
the results of the decomposition of the ngc3281 spectra are shown in fig .
[ pahfit ] , and the derived silicate optical depths are given in tab .
[ lines ] .
clearly , @xmath40 varies along the slit , being @xmath40 = [email protected] at the unresolved nucleus and @xmath40 = [email protected] at 130pc nw extraction . in constrast
, it reaches @xmath40 = [email protected] for the 130pc se extraction .
the dust extinction ( a@xmath2 ) values estimated using the pahfit derived line of sight depth , column 10 in tab .
[ lines ] , are much larger than those obtained from the s@xmath34 .
the latter are obtained from the depth at 9.7 while the pahfit code takes the whole profile of the silicate absorption feature into account , thus the s@xmath34 indicator should be taken as a lower limit for the a@xmath2 .
the obtained opacity values were compared to those of the seyfert sample from @xcite .
the histogram of @xmath40 ( fig .
[ histogram ] ) shows that the dust opacity in ngc3281 is one of the highest observed in seyfert galaxies .
the thermal dust continuum components required to fit the observed spectra of the unresolved nucleus and 130pc se is a black - body with t = 300k , while the 130pc nw spectrum requires it a black - body of t = 200k .
this colder region has the deepest silicate absorption , consequently is more heavily obscured ( a@xmath2 = 102@xmath1826mag ) than the nucleus and the se region .
the presence of this obscuring material is consistent with the spatial distribution of the optical reddening found by @xcite .
note that our slit position is perpendicular to the cone observed in [ oiii ] .
see their fig . 9 and 2 .
in order to get additional support of that highly reddened region , we overplot the t - recs slit position on the [ oiii]@xmath415007 maps taken from @xcite .
we found that the 130pc nw region coincide with a region where [ oiii ] emission is not observed in agreement with those previously found from x - ray data .
that is , the ngc3281 is a heavily obscured source .
there are several torus models , some authors take it as having an uniform dust density distribution ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , for dust grains to survive in the torus environment , they should be formed by clumpy structures @xcite , and new models have been developed where the dust grains are distributed in a clumpy medium ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , we compare the uncontaminated nuclear spectrum ( see below ) with theoretical sed obtained from nenkova s models ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in their modeling , nenkova et al .
assume that the dusty material surrounding the agn is distributed in clumps forming a toroidal geometry , and determine the following parameters : _ i _ ) the number of clouds _
@xmath42 _ in the torus equatorial radius ; _ ii _ ) the optical depth for an individual cloud in the v band ( @xmath4 ) ; _ iii _ ) the radial extension of the clumpy distribution , @xmath43 , where @xmath44 and @xmath45 are the outer and inner radius of the torus , respectively ; _ iv _ ) the power law index @xmath46 that describes the radial density profile ( @xmath47 ) ; _ v _ ) the torus angular width , constrained by a gaussian angular distribution of @xmath48 ; _ vi _ ) the observed viewing angle @xmath5 . in order to properly analyze the spectrum from the nuclear dusty structure ,
it is necessary to first remove the host galaxy contribution . for this , we have subtracted from the central aperture the spectrum obtained from averaging the two adjacent extractions ( apertures -1 and 1 , fig .
[ fenda ] ) .
noise effects are reduced by smoothing ( with a rectangular box of size 5 pixels ) the final spectrum .
the emission lines were removed by simple interpolation and the region where the telluric band is located ( fig.[spectra ] ) were not used in the fit .
then , the smoothed spectrum was compared with the @xmath49 clumpy theoretical seds , and the best fit is obtained by searching for the minimum in the equation : @xmath50 where n is the number of data points in the spectrum , f@xmath51 , f@xmath52 and @xmath53 are the observed , theoretical fluxes at each wavelength and their respective uncertainties .
the later were taken as being 10% of @xmath54 , which are the expected ones for t - recs observations @xcite .
note that both f@xmath51 and f@xmath52 were normalized to unit at 9.0 , with the uncertainties being properly propagated .
the parameter set that provides the best fit is shown in table [ statistic ] , and the corresponding theoretical sed is over - plotted on the ngc3281 decontaminated nuclear spectrum ( fig . [ nenkova ] ) . due to the very large number of models required to cover the parameter space
, we estimate the uncertainties for the best model following the approach of ( * ? ? ?
* see also @xcite ) , but instead of using only the three best models , we calculated the mean and standard deviation for the parameters considering all models with a @xmath55 value within 10% of the best - fitting result ( 76 in total ) .
the locus of the corresponding theoretical seds is plotted as a grey region in figure [ nenkova ] .
our approach is similar to that used by @xcite , thus , we have also tested other @xmath55 deviation fractions ( 5% , 10% , 15% and 20% , corresponding to 17 , 76 , 151 and 291 acceptable solutions , respectively ) . in general
our technique produce similar results than that obtained by @xcite , however , our methodology tends to produce a locus with less acceptable solutions than nikutta s one .
the mean parameters derived with the different @xmath55 deviations are shown in table [ statistic ] .
the physical parameters of the best - fit model to the nuclear uncontaminated seds suggest that ngc3281 hosts a dusty toroidal structure .
the dusty clouds , each with an optical depth @xmath4 = 40mag , occupy a toroidal volume with @xmath3 = 20 .
the distribution of the clouds follows a power - law radial behavior ( r@xmath56 ) , and the number of clouds along the equatorial radius is 10 .
the angular distribution of the clouds is characterized by a width @xmath48 = 70@xmath6 .
these results are in agreement with the unified agn model , which requires the presence of a dusty structure that obscures the broad line region in sy 2 galaxies . from the fitted models we would be looking in a direction nearby of the torus equator ( @xmath5 = 60@xmath6 ) in general , the torus parameters derived here are similar to those obtained by @xcite , except for the number of clouds and optical depth per single cloud , where they find only 5 clouds with @xmath4 = 10 in the equatorial radius
. this discrepancy can be due to the different constraints available , as @xcite used mid - ir photometric data ( see also * ? ? ?
* ; * ? ? ?
* ) , while the data presented here would allow for a more detailed description of the spectral behavior of the silicate dust , thus providing a better input to the modeling procedure .
the @xcite models adopt a standard dust - to - gas ratio , and assume the column density of a single cloud to be n@xmath57 .
in addition , the total number of clouds at a viewing angle @xmath5 is given by a gaussian distribution : n@xmath58 = n@xmath29exp@xmath59 $ ] , thus resulting in a column density of n@xmath60 = _ _ n@xmath58__n@xmath23 at the observer view angle .
using the above , we can then derive a column density of n@xmath61 for the uncontaminated nuclear spectrum of ngc3281 , which is consistent with the values derived by @xcite from x - ray observations ( @xmath62 ) .
moreover , @xcite demonstrated that for heavily obscured agn the iron k@xmath10 ew values depend on the torus geometry .
we compare the torus geometrical parameters derived in our paper with those of @xcite , and found that the present torus aperture ( @xmath48 = 70@xmath6 ) correspond to @xmath63 = 20@xmath6 in the @xcite s models .
the observer view angle has the same definition for both models ( i = 60@xmath6 ) . from figure 2 of @xcite and using the above parameters we estimate that the iron k@xmath10 ew for ngc3281 is @xmath11 2 - 3kev , which is in agreement with that found by @xcite , confirming their results , that this galaxy would shows a large ew for the iron k@xmath10 line .
we point out that this torus geometry was inferred from the silicate absorption line , this indicates that the x - ray absorbing column density , which makes ngc3281 a compton - thick source , may also be responsible for the absorption at 9.7@xmath0 m .
therefore , our results provide strong evidence that the silicate dust responsible for such absorption is located in the agn torus .
following @xcite , and adopting a dust sublimation temperature of 1500k , we estimate a torus outer radius of r@xmath7 11pc .
this value is consistent with the results of @xcite and @xcite .
we also derive the intrinsic bolometric luminosity for the agn from the best - fit model , finding l@xmath64 = 1.9@xmath65erg s@xmath16 .
this value agrees with the intrinsic luminosity obtained by @xcite from x - ray data , when applying the conversion factor suggested by @xcite , resulting l@xmath66
@xmath67erg s@xmath16 .
we note that ngc3281 has a high x - ray luminosity when compared to objects studied by @xcite .
in this work we present high - spatial resolution mid - ir spectra of the compton - thick sy 2 galaxy ngc3281 .
the ground - based mid - ir data were taken with the t - recs spectrograph , and our main results can be summarized as follows .
* ngc3281 s spectra show a very deep silicate absorption , as well as [ siv]10.5@xmath0 m and [ neii]12.7@xmath0 m ionic lines
. however , the spectra do not present pah emission bands , which makes this galaxy uncommon among the sy 2s class @xcite .
furthermore , the dust is concentrated inside a radius of 200pc . *
the optical extinction , a@xmath2 , of the nucleus and its neighborhood was estimated using the s@xmath34 indicator and pahfit code . from the s@xmath34 indicator we infer the nuclear dust extinction a@xmath38 = 24@xmath182mag , while the pahfit code provides a@xmath2 = 83@xmath1822mag .
we point out that the first value can only be used as a lower limit .
this large extinction confirms that this galaxy , if compared with @xcite seyfert sample ( 83 sources ) , is a heavily obscured source compatible with a compton - thick galaxy scenario . *
the black - body temperature due to dust continuum components for the unresolved nucleus and the region at 130pc se is t = 300k .
a colder black - body ( t = 200k ) was found for the region at 130pc nw , which also has the highest derived visual extinction ( a@xmath2 = [email protected] ) .
* the nuclear uncontaminated spectrum follows a clumpy torus model @xcite , which suggests that ngc3281 has a dusty toroidal structure .
the dusty clouds occupy a toroidal volume with @xmath3 = 20 , with each cloud having an optical depth @xmath4 = 40mag .
the distribution of the clouds follows a power - law radial behavior @xmath68 and the number of clouds in the equatorial radius is 14 .
the angular distribution of the clouds has width of @xmath48 = 70@xmath6 .
these results agree with the unified agn model that to block the broad line region emission of sy 2 galaxies , requires a dusty structure . according to the best - fit model
, we would be looking in the direction nearby of the torus equatorial radius ( @xmath5 = 60@xmath6 ) , having an outer radius of r@xmath7 11pc .
* we derive a column density of n@xmath23 = 1.2@xmath9 for ngc3281 indicates that the x - ray absorbing column density , which classifies ngc3281 as a compton - thick source , may also be responsible for the absorption at 9.7@xmath0 m .
in addition , our results provide strong evidence that the silicate dust responsible for such absorption is located in the agn torus . from nenkova s models we find that the bolometric luminosity of this agn is l@xmath64 = 1.9@xmath65erg s@xmath16 , which shows that ngc3281 is a very luminous source .
lccccccccc 130pc se & -1 & [email protected] & 0.7 & - & - & [email protected] & 26@xmath185 & [email protected] & 102@xmath1826 + center & 0 & [email protected] & 0.1 & [email protected] & 0.02 & [email protected] & 24@xmath185 & [email protected] & 83@xmath1822 + 130pc nw & 1 & [email protected] & 0.2 & [email protected] & 0.06 & [email protected] & 28@xmath188 & [email protected] & 85@xmath1822 + 195pc nw & 2 & - & - & - & - & [email protected] & 21@xmath184 & - & - + lccccc torus angular width ( @xmath48 ) & 70@xmath6 & 68@xmath6@xmath183@xmath6 & 62@xmath6@xmath186@xmath6 & 62@xmath6@xmath187@xmath6 & 61@xmath6@xmath188@xmath6 + radial extent of the torus ( @xmath69 ) & 20 & 20@xmath180 & 23@xmath184 & 28@xmath1818&38@xmath1827 + number of clouds in the torus equatorial radius ( @xmath70 ) & 14 & 13@xmath181 & 13@xmath182&13@xmath182&13@xmath182 + power - law index of the radial density profile ( @xmath46 ) & 1.5 & 1.5@xmath180 & [email protected]&[email protected]&[email protected] + observers viewing angle ( @xmath5 ) & 60@xmath6 & 69@xmath6@xmath1811@xmath6 & 76@xmath6@xmath1811@xmath6&76@xmath6@xmath1812@xmath6 & 72@xmath6@xmath1818@xmath6 + optical depth for individual cloud ( @xmath71 ) & 40mag & 40@xmath180mag & 40@xmath180mag&40@xmath183mag&41@xmath188mag + total number of solution & & 17 & 76 & 151 & 291 + we thank an anonymous referee for helpful suggestions . based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , ministrio da cincia e tecnologia ( brazil ) and ministerio de ciencia , tecnologa e innovacin productiva ( argentina ) .
we thank e. balbinot , as well as charles bonatto , and a.c.c .
acknowledges support from cnpq ( grant 308985/2009 - 5 ) e fapesp ( grant 2010/19029 - 0 ) .
alonso - herrero , a. , ramos almeida , c. , mason , r. , et al .
2011 , , accepted antonucci , r. 1993 , , 31 , 473 bohlin , r. c. , savage , b. d. , & drake , j. f. 1978 , , 224 , 132 cohen , m. , walker , r. g. , carter , b. , hammersley , p. , kidger , m. , & noguchi , k. 1999 , , 117 , 1864 , durret , f. & bergeron , j. 1988 , , 75 , 273 draine , b. t. 2003 , , 41 , 241 elvis , m. , risaliti , g. , nicastro , f. , miller , j. m. , fiore , f. , & puccetti , s. 2004 , , 615 , l25 fritz , j. , franceschini , a. , & hatziminaoglou , e. 2006 , , 366 , 767 gallimore , j. f. , yzaguirre , a. , jakoboski , j. , et al .
2010 , , 187 , 172 granato , g. l. , danese , l. , & franceschini , a. 1997 , , 486 , 147 hao , l. , weedman , d. w. , spoon , h. w. w. , marshall , j. a. , levenson , n. a. , elitzur , m. , & houck , j. r. 2007 , , 655 , l77 hnig , s. f. , beckert , t. , ohnaka , k. , & weigelt , g. 2006 , , 452 , 459 houck , j. r. , roellig , t. l. , van cleve , j. , et al .
2004 , , 154 , 18 jaffe , w. , meisenheimer , k. , rttgering , h. j. a. , et al .
2004 , , 429 , 47 kemper , f. , vriend , w. j. , & tielens , a. g. g. m. 2004 , , 609 , 826 krolik , j. h. , & begelman , m. c. 1988 , , 329 , 702 levenson , n. a. , krolik , j. h. , ycki , p. t. , heckman , t. m. , weaver , k. a. , awaki , h. , & terashima , y. 2002 , , 573 , 81l levenson , n. a. , sirocky , m. m. , hao , l. , spoon , h. w. w. , marshall , j. a. , elitzur , m. , & houck , j. r. 2007 , , 654 , 45l marshall , n. , warwick , r. s. , & pounds , k. a. 1981 , , 194 , 987 mason , r. e. , geballe , t. r. , packham , c. , levenson , n. a. , elitzur , m. , fisher , r. s. , & perlman , e. 2006 , , 640 , 612 mason , r. e. , levenson , n. a. , shi , y. , packham , c. , gorjian , v. , cleary , k. , rhee , j. , & werner , m. 2009 , , 693 , 136l mulchaey , j. s. , koratkar , a. , ward , m. j. , wilson , a. s. , whittle , m. , antonucci , r. r. j. , kinney , a. l. , & hurt , t. 1994 , , 436 , 586 nenkova , m. , ivezi , .
, & elitzur , m. 2002 , , 570 , l9 nenkova , m. , sirocky , m. m. , ivezi , . , & elitzur , m. 2008 , , 685 , 147 nenkova , m. , sirocky , m. m. , nikutta , r. , ivezi , . , & elitzur , m. 2008 , , 685 , 160 nenkova , m. , sirocky , m. m. , nikutta , r. , ivezi , . , & elitzur , m. 2010 , , 723 , 1827 nikutta , r. , elitzur , m. , & lacy , m. 2009 , , 707 , 1550 phillips , m. m. , charles , p. a. , & baldwin , j. a. 1983 , , 266 , 485 pier , e. a. , & krolik , j. h. 1992 , , 401 , 99 radomski , j. t. , pia , r. k. , packham , c. , telesco , c. m. , & tadhunter , c. n. 2002 , , 566 , 675 ramos almeida , c. , levenson , n. a. , rodrguez espinosa , j. m. , et al .
2009 , , 702 , 1127 ramos almeida , c. , levenson , n. a. , alonso - herrero , a. , et al .
2011 , , 731 , 92 roche , p. f. , packham , c. , aitken , d. k. , & mason , r. e. 2007 , , 375 , 99 sales , dinalva a. , pastoriza , m. g. , & riffel , r. 2010 , , 725 , 605 sandage , a. 1978 , , 83 , 904 sandage , a. , & brucato , r. 1978 , , 84 , 472 sanders , d. b. , mazzarella , j. m. , kim , d. c. , surace , j. a. , & soifer , b. t. 2003 , , 126 , 1607 schmitt , h. r. , ulvestad , j. s. , antonucci , r. r. j. , & kinney , a. l. 2003 , , 132 , 199 schmitt , h. r. , donley , j. l. , antonucci , r. r. j. , hutchings , j. b. & kinney , a. l. 2003 , , 148 , 327 siebenmorgen , r. , krgel , e. , & spoon , h. w. w. 2004 , , 414 , 123 simpson , c. 1998 , , 509 , 653 smith , j. d. t. , draine b. t. , dale d. a. , et al .
2007 , , 656 , 770 spoon , h. w. w. , marshall , j. a. , houck , j. r. , elitzur , m. , hao , l. , armus , l. , brandl , b. r. , & charmandaris , v. 2007 , , 654 , 49 storchi - bergmann , t. , wilson , a. s. , & baldwin , j. a. 1992 , , 396 , 45 sturm , e. , lutz , d. , tran , d. , feuchtgruber , h. , genzel , r. , kunze , d. , moorwood , a. f. m. , & thornley , m. d. 2000 , , 358 , 481 telesco , c. m. , pi~ na , robert k. , hanna , kevin t. , julian , jeffrey a. , hon , david b. , & kisko , thomas m. 1998 , proc .
spie , 3354 , 534 tristram , k. r. w. , meisenheimer , k. , jaffe , w. , et al .
2007 , , 474 , 837 vignali , c. , & comastri , a. 2002 , , 381 , 834 young , s. , packham , c. , mason , r. e. , radomski , j. t. & telesco , c. m. 2007 , , 378 , 888 weedman , d. w. , hao , lei , higdon , s. j. u. , et al .
2005 , , 633 , 706 wu , y. , charmandaris , v. , huang , j. , spinoglio , l. , & tommasin , s. 2009 , , 701 , 658
|
we present mid infrared ( mid - ir ) spectra of the compton - thick seyfert 2 galaxy ngc3281 , obtained with the thermal - region camera spectrograph ( t - recs ) at the gemini south telescope .
the spectra present a very deep silicate absorption at 9.7@xmath0 m , and [ siv]10.5@xmath0 m and [ neii]12.7@xmath0 m ionic lines , but no evidence of pah emission .
we find that the nuclear optical extinction is in the range 24 @xmath1 a@xmath2 @xmath1 83mag .
a temperature t = 300k was found for the black - body dust continuum component of the unresolved 65pc nucleus and at 130pc se , while the region at 130pc reveals a colder temperature ( 200k ) .
we describe the nuclear spectrum of ngc3281 using a clumpy torus model that suggests that the nucleus of this galaxy hosts a dusty toroidal structure .
according to this model , the ratio between the inner and outer radius of the torus in ngc3281 is @xmath3 = 20 , with 14 clouds in the equatorial radius with optical depth of @xmath4 = 40mag .
we would be looking in the direction of the torus equatorial radius ( @xmath5 = 60@xmath6 ) , which has outer radius of r@xmath7 11pc .
the column density is [email protected]@xmath9 and iron k@xmath10 equivalent width ( @xmath11 0.5 - 1.2kev ) are used to check the torus geometry .
our findings indicate that the x - ray absorbing column density , which classifies ngc3281 as a compton - thick source , may also be responsible for the absorption at 9.7@xmath0 m providing strong evidence that the silicate dust responsible for this absorption can be located in the agn torus .
|
the functional schrdinger picture is based on the projection of quantum states and operators on the field amplitude basis .
guth and pi @xcite have already used it in the study of inflationary perturbations and related the width of the gaussian vacuum wave functional in de sitter spacetime to heisenberg picture scalar fields , thus casting quantum field theory in terms familiar from non - relativistic quantum mechanics .
a kernel is a quantum mechanical evolution operator in the functional schrdinger picture and is the fundamental object of any quantum mechanical theory .
it represents a transition amplitude from an arbitrary initial state at time @xmath0 to an arbitrary final state at time @xmath1 .
moreover , it allows to calculate the forward time evolution of any given initial field configuration . in this paper
we demonstrate that the kernel can be expressed solely in terms of the causal propagator and derivatives of the causal propagator thus elaborating further on the connection with the heisenberg ( operator ) picture for quantum field theories .
furthermore , the appearance of the causal propagator makes the causality structure of the theory evident .
the kernel allows for many physical applications in quantum field theory .
for example , if interactions between various scalar fields or fermionic fields are linear in ( one of the ) matter fields , the formalism developed here can straightforwardly be applied .
furthermore , another natural application of the kernel can be found in the study of decoherence @xcite accounting for the quantum - to - classical transition in the early universe .
after having reviewed some basic quantum mechanics in the current section , we will calculate the kernel for a free scalar field in section [ the kernel in the non - interacting case ] , and a scalar field coupled to a source term in section [ the kernel in the interacting case ] in cosmological spacetimes ( spatially homogeneous backgrounds ) . in section [ causality in quantum mechanics ]
we elaborate on causality in quantum field theory expressed in the functional schrdinger picture and end in section [ applications ] with some examples , to wit , the simple harmonic oscillator , the kernel in de sitter spacetime and the time evolution of the vacuum state in friedmann - lematre - robertson - walker or flrw spacetimes .
let us begin by recalling some basic identities from quantum mechanics .
a quantum state @xmath2 at time @xmath1 can in general be expressed in terms of the evolution operator @xmath3 and the quantum state @xmath4 at initial time @xmath0 as follows : latexmath:[\[\label{evolution } @xmath2 , the evolution operator @xmath3 obeys the schrdinger equation : @xmath6 where @xmath7 is the hamiltonian operator of the system under consideration .
the formal solution is a time ordered ( @xmath8 ) exponential : @xmath9 note that @xmath10 is implied and when @xmath11 time ordering should be replaced with anti - time ordering ( @xmath12 ) . from this equation
we can easily infer some important properties the evolution operator satisfies : [ evolutionproperties ] @xmath13 let us now state a relation between quantum mechanical expectation values and commutators ( see for example @xcite ) which will play an important rle in the interpretation of the kernel .
we consider the expectation value of some general operator @xmath14 in the schrdinger picture : @xmath15 now , by using ( [ evolution : solution ] ) for the evolution operator : @xmath16 one can prove by induction that at each order in @xmath7 one has : @xmath17 \cdots\right]\right]\right|\psi(t')\right\rangle.\ ] ] we can conclude that expectation values , naturally defined in terms of evolution operators as in ( [ expectationvalueq2 ] ) , can in a fully equivalent manner be calculated from the expectation value of nested commutators , as in ( [ expectationvalueq3 ] ) .
information about observables is hence equally well stored in a series of commutators as in evolution operators . in quantum mechanics causality
corresponds to the statement that all commutators of observables , including non - commuting observables , vanish outside past and future light cones .
alternatively , measurements performed at points in spacetime with a spacelike separation can be carried out simultaneously .
this equivalence relation already suggests that quantum mechanics is a causal theory in full generality since , up to higher order irreducible @xmath18-point functions , expectation values can be expressed in terms of the commutator , a manifestly causal quantity .
these are vital observations , and we will return to them shortly . given some quantum field operator @xmath19 in the heisenberg picture one can construct various vacuum expectation values by means of the schwinger - keldysh formalism ( see e.g. @xcite ) .
we will be interested in a particular linear combination of the two wightman functions ( @xmath20 and @xmath21 ) to construct what we will henceforth refer to as the causal propagator : @xmath22\right| \omega\right\rangle \equiv \left\langle\omega\left|\hat{\phi}(x ) \hat{\phi}(x ' ) - \hat{\phi}(x ' ) \hat{\phi}(x ) \right|\omega\right\rangle \,.\ ] ] in particular , if we assume a spatially homogeneous background and a quadratic hamiltonian , expanding in terms of creation and annihilation operators : @xmath23 yields : @xmath24 where : @xmath25 here , the @xmath26 originates from imposing the standard commutation relations between creation and annihilation operators .
furthermore , note the field modes are homogeneous , i.e. : @xmath27 depends on @xmath28 .
this propagator is causal in the quantum mechanical sense because it originates from the commutator @xcite , unlike for example the feynman or ( anti-)time ordered propagators . in the functional schrdinger picture
@xcite a quantum mechanical state @xmath2 is realised by a wave functional @xmath29 which is a functional of the @xmath30-number functions @xmath31 defined by the projection on the field amplitude basis @xmath32 , where @xmath33 . the action of a quantum field operator @xmath34 and its associated canonical momentum @xmath35 are given by : @xmath36
let us now examine the quantum mechanical evolution operator ( [ evolution : solution ] ) in the functional schrdinger picture : @xmath37\right ] \,.\ ] ] in the above equation , @xmath38 is the so - called kernel , a transition amplitude from some initial state @xmath39 at @xmath0 to the state @xmath40 at @xmath1 . the kernel can be expressed in terms of a path integral of the action where we integrate over all intermediate field configurations .
it is interesting to note that from properties ( [ evolutionproperties ] ) we deduce the following symmetry requirements for @xmath41 : [ symreqk ] @xmath42 where the functional delta function has to be understood as : @xmath43 .
the action for a real scalar field @xmath44 for a finite time interval ranging between @xmath0 and @xmath1 generally valid for real quantum fields in curved spacetimes is given by : @xmath45 = \int d^{4}x \ , \sqrt{-g } \left ( -\frac{1}{2}\ , \partial_{\alpha } \phi(x ) \
, \partial_{\beta } \phi(x ) \ , g^{\alpha\beta } -\frac{1}{2}\,(m^{2}+\xi r)\phi^{2}(x)\right ) \,,\ ] ] where @xmath46 denotes the ricci curvature scalar and @xmath47 $ ] .
let us specialise to flrw spacetimes in which the metric is given by @xmath48 where @xmath49 is the scale factor of the universe .
classically , this action leads to the standard equation of motion : @xmath50 where @xmath51 is the scalar dalambertian .
we write : @xmath52 and insert this into equation ( [ action ] ) .
the boundary conditions on @xmath44 are carried by the classical field only , i.e. : @xmath53 and @xmath54 , and straightforwardly result into the requirement that @xmath55 . assuming that the classical field ( [ motionphi ] ) vanishes at spatial infinity , two straightforward partial integrations yield : @xmath56 = s[\delta\phi ] + s[\phi_{\mathrm{cl } } ] = s[\delta\phi ] + \frac{1}{2 } \int d^{3}\mathbf{x } \,\phi_{\mathrm{cl}}(x)\ , \pi_{\mathrm{cl}}(x ) \big|_{t'}^{t } \,,\ ] ] where we have recognised the canonical momentum associated to @xmath57 given by @xmath58 .
equation ( [ projection - functionalintegral ] ) boils down to the following result : @xmath59 \int_{\delta\phi(t')=0}^{\delta\phi(t)=0 } \mathcal{d}\delta\phi\,\,\exp\left [ \frac{i}{\hbar}\,s[\delta\phi]\right ] \,.\ ] ] note that the remaining path integral represents the transition amplitude between zero field states and is thus independent of the original boundary conditions on the field @xmath44 .
hence we can absorb this factor into the overall normalisation of the kernel : @xmath60\,.\ ] ] next , we fourier expand the classical field : @xmath61 which is well defined because we consider a spatially flat flrw background .
note the property @xmath62 and the omission of the subscript cl " of the field modes for future convenience .
we must keep in mind though , that also the field modes are classical in the sense that they obey the fourier transform of equation of motion ( [ motionphi ] ) : @xmath63 where @xmath64 is the hubble parameter .
we thus have : @xmath65 = \left.\frac{1}{4 } \ , \int \frac{d^{3}\mathbf{k}}{(2\pi)^{3}}\ , \left[a^{3}(\tilde{t}\ , ) \partial_{\tilde{t}}\left | \phi_{\mathbf{k } } ( \tilde{t}\,)\right running time @xmath66 is defined on the interval @xmath67 .
let us denote the two fundamental solutions of ( [ motionphi2 ] ) by @xmath68 and @xmath69 .
in general , @xmath70 is a linear superposition of the two fundamental solutions , i.e. : @xmath71 .
we are now in the position to impose the boundary conditions at initial and final times and solve for @xmath72 and @xmath73 .
we thus find : @xmath74 where @xmath75 is related to the causal propagator ( [ commutatorpropagatorf ] ) as follows : @xmath76 where @xmath77 is a unit vector in k - space normal to an arbitrary plane through the origin .
the sign - function is introduced in order to preserve the odd symmetry under @xmath78 .
one can thus express the solution to differential equation ( [ motionphi2 ] ) in terms of its boundary conditions and causal propagators exclusively .
we believe that this is also true for general spacetimes and fields of non - zero spin @xcite .
note that the wronskian yields : @xmath79 \equiv \phi_{\mathbf{k}}(t ) \dot{\phi}^{\ast}_{\mathbf{k}}(t ) - \phi^{\ast}_{\mathbf{k}}(t ) \dot{\phi}_{\mathbf{k}}(t ) = \left\{|\alpha_{\mathbf{k}}|^{2}- |\beta_{\mathbf{k}}|^{2}\right\ } w[\chi_{\mathbf{k}}(t ) , \chi_{\mathbf{k}}^{\ast}(t)]\,,\ ] ] where : @xmath80 generally , this equation does not equal unity , a requirement usually imposed for a consistent canonical quantisation in the heisenberg picture , because the boundaries are arbitrary .
indeed , this is an important difference between the schrdinger and heisenberg pictures .
next , we substitute ( [ phisolutionpropagator ] ) into our modified action , equation ( [ actionrewrite2 ] ) . the finite volume ( in position space )
result for the kernel in fourier space reads : @xmath81 \,,\ ] ] where @xmath82 is a normalisation constant and where : [ propkernelconst ] @xmath83 + a^{3}(t ' ) w[\chi_{\mathbf{k}}(t ' ) , \chi_{\mathbf{k}}^{\ast}(t ' ) ] \big\ } \label{propkernelconstd}\,.\end{aligned}\ ] ] note that @xmath84 due to the homogeneity of space .
the volume factor appearing in ( [ propkernel ] ) is due to the identification @xmath85 , valid in the infinite volume limit , which we use throughout this paper . in flrw spacetimes ,
the wronskian equals : @xmath86 = \frac{i}{a^{3}(t ) } \mathrm{sgn}(\mathbf{k}\cdot\hat{\mathbf{n } } ) \,.\ ] ] note again the appearance of the sign - function .
finally , the appropriate normalisation constant can be determined upon inserting the kernel ( [ propkernel][propkernelconst ] ) into the functional schrdinger equation : @xmath87 we consider the following general form for the kernel ( [ propkernel ] ) in position space : @xmath88 \,,\ ] ] where we define for example @xmath89 for brevity and where : @xmath90 and similarly for @xmath91 and @xmath92 . using ( [ propkernelconst ] ) , we can easily check that the fourier transform of ( [ kernelpositionspace ] ) satisfies the functional schrdinger equation at each ( non - zero ) order in the fields @xmath40 and @xmath39 . at zeroth order ,
the functional schrdinger equation reads : @xmath93 after switching to fourier space , we write @xmath94 .
next , it is a non - trivial step to exploit the functional schrdinger equation at order @xmath95 in order to explicitly preserve the invariance under @xmath78 .
this yields : @xmath96 we can solve this differential equation straightforwardly : @xmath97 where the time independent constant @xmath98 has to be fixed by condition ( [ symreqk2 ] ) , or , alternatively by requiring @xmath99 .
the result is : @xmath100 .
hence , the normalisation constant is given by : @xmath101 finally , we can straightforwardly check by substitution that symmetry requirement ( [ symreqk1 ] ) is also met .
the kernel we have just constructed can readily be applied to many applications when dealing with non - interacting scalar fields in homogeneous spacetimes , as we will come to discuss in section [ applications ] .
so far we have only examined the kernel for a free , non - interacting , quantum field .
it is thus natural to turn to the interacting case which is particularly interesting for many physical situations .
we will generalise the action in equation ( [ action ] ) and incorporate a spacetime dependent source term @xmath102 coupled linearly to @xmath44 .
this particular type of interaction can model interactions with other scalar or fermionic fields , provided it is linear in one of the scalar fields .
the action is given by : @xmath103 = \int d^{4}x \ , \sqrt{-g } \left ( -\frac{1}{2}\ , \partial_{\alpha } \phi(x ) \
, \partial_{\beta } \phi(x ) \ , g^{\alpha\beta } - \frac{1}{2 } \,\left ( m^{2 } + \xi r\right ) \phi^{2}(x ) + j(x)\phi(x ) \right ) \,,\ ] ] where , again
, the time integral is performed over the finite interval from @xmath0 to @xmath1 and the position integral is over all space .
this action leads to the following equation of motion for the classical field : @xmath104 fourier transforming and using the flrw metric as before yields : @xmath105 analogously to equation ( [ motionphi2 ] ) .
we split the quantum field as in ( [ qfielcfield ] ) and require that the boundary conditions in the path integral in ( [ projection - functionalintegral ] ) are carried solely by the classical field .
we thus arrive at : @xmath106 = s_{0}[\delta\phi ] + \frac{1}{2 } \int d^{3}\mathbf{x}\ , \phi_{\mathrm{cl}}(x ) \ , \pi_{\mathrm{cl}}(x ) \big|_{t'}^{t } + \frac{1}{2 } \int d^{4}x \sqrt{-g}\ \phi_{\mathrm{cl}}(x ) j(x)\,.\ ] ] first of all note that @xmath107 $ ] refers to the contribution to the action of @xmath108 in the absence of interactions . secondly ,
when comparing to the non - interacting case , we see that the second term in the equation above is unchanged and the source enters only through the third term .
analogous to equation ( [ actionrewrite2 ] ) the equation above in fourier space is given by : @xmath109 = s_{0}[\delta\phi ] + \left.\frac{1}{4 } \
, \int \frac{d^{3}\mathbf{k}}{(2\pi)^{3}}\ , \left [ a^{3}(\tilde{t}\ , ) \partial_{\tilde{t}}\left | \phi_{\mathbf{k } } ( \tilde{t}\,)\right |^{2}\right ] \right|_{\tilde{t}=t'}^{\tilde{t}=t } + \frac{1}{2}\int \frac{d^{3}\mathbf{k}}{(2\pi)^{3}}\int_{t'}^{t}d\tilde{t}\,a^{3}(\tilde{t}\ , ) j_{\mathbf{k}}^{\ast}(\tilde{t}\,)\ , \phi_{\mathbf{k}}(\tilde{t}\ , ) \,.\ ] ] the homogeneous solution of ( [ motionphi3s ] ) is given by ( [ phisolutionpropagator ] ) .
employing the green s function method and noting that only the homogeneous solution carries the boundary conditions , we find the following total solution : @xmath110 where : @xmath111 and where @xmath112 is the appropriate green s function corresponding to ( [ motionphi3s ] ) : @xmath113 finally , we easily derive the symmetry relation @xmath114 . upon recalling the definition of the wronskian , @xmath115_{t'\rightarrow t}$ ] , we see that the green s function ( [ sourcesolution ] ) is , as expected , expressed solely in terms of the causal propagator .
we proceed completely analogously by substituting expansion ( [ solutionlincombs1 ] ) into the action , equation ( [ actionrewrite2s ] ) , collecting all terms at each order , i.e. : @xmath116 , @xmath117 , @xmath118 and @xmath119 at quadratic order and @xmath120 , @xmath121 , @xmath122 and @xmath123 as the linear contributions .
the source contributes through the linear terms only and the quadratic ones remain unaffected . furthermore , we can safely omit the terms at zeroth order in the fields , because we will incorporate those into the overall normalisation .
hence the full result for the kernel reads : @xmath124 \nonumber\,,\end{aligned}\ ] ] where @xmath82 is again a normalisation constant and where the @xmath125- , @xmath126- and @xmath127-functions are given by ( [ propkernelconst ] ) and , finally , where : @xmath128 \label{propkernelconstse } \\ f(\mathbf{k},t , t ' ) & = & \frac{i}{2\hbar}\bigg [ a^{3}(t')\partial_{t ' } s^{\ast}_{\mathbf{k}}(t')- \int_{t'}^{t}d\tilde{t } \ , a^{3}(\tilde{t}\ , ) j^{\ast}_{\mathbf{k}}(\tilde{t}\ , ) \frac{g_{\mathbf{k}}(t,\tilde{t } ) } { g_{\mathbf{k}}(t , t ' ) } \bigg ] \label{propkernelconstsf } \,.\end{aligned}\ ] ] for a detailed derivation see @xcite .
the derivative @xmath129 should be interpreted as @xmath130 .
since @xmath131 , note that the contribution at order @xmath121 carries the opposite sign as compared to the contribution at order @xmath120 which _ mutatis mutandis _ holds for @xmath132 and @xmath133 . let us now find the new normalisation constant .
the functional schrdinger equation changes in comparison with ( [ funcschrodingereqn ] ) to : @xmath134 the position space form for the kernel ( [ propkernels ] ) generalises equation ( [ kernelpositionspace ] ) to : @xmath135\,.\end{aligned}\ ] ] substitution into the functional schrdinger equation ( [ funcschrodingereqns ] ) leads again to a number of equations at various orders in the fields @xmath40 and @xmath39 that are indeed satisfied simultaneously .
writing @xmath136 , the zeroth order equation in fourier space reads : @xmath137 we can solve this equation straightforwardly by : @xmath138 \,,\ ] ] where @xmath139 is again a time independent constant .
note that the lower boundary of the integral is @xmath0 because our solution for the kernel is constructed such that it vanishes for times less than @xmath0 .
finally , @xmath139 has to be fixed by condition ( [ symreqk2 ] ) as before .
now , in the limit when @xmath140 the kernel can easily be verified to be : @xmath141 \nonumber\,.\end{aligned}\ ] ] the dominant terms in the exponent behave as @xmath142 as @xmath143 and , consequently , the terms involving the source all give a vanishing contribution in this limit .
equivalently , the shift of the delta - function induced by the source vanishes at the lower boundary .
we conclude therefore that ( [ propkernellimits ] ) indeed reduces to a representation of a delta - function correctly .
the result for @xmath139 is hence unchanged compared to the free case ( [ propkernelconstm ] ) .
furthermore , note that the width of the gaussian in the limit when @xmath143 behaves as @xmath144 , which indicates an early time diffusive wave packet spreading in configuration space . finally , it can easily be verified that the obtained solution satisfies the correct symmetry properties ( [ symreqk1 ] ) .
let us now examine the causal structure of quantum field theory in the formalism developed above . in this section we will prove that a ) the causal propagator ( [ commutatorpropagator ] ) in position space vanishes when @xmath145 in de sitter space and b ) the same light cone structure is also present in the functional schrdinger picture and can be inferred from the kernel we have constructed . here
@xmath146 denotes conformal time defined by @xmath147 .
the clear geometrical interpretation of causality in position space can not be easily transferred to fourier space because a fourier transformation is nonlocal : one can not simply draw `` light cones '' in fourier space .
a delicate cancellation occurs such that a superposition of fourier modes cancel precisely outside past and future light cones .
since flrw spacetimes are conformal , their causality structure is most easily described in conformal coordinates , in which the light cones are simply @xmath148 . in de sitter space
the scale factor @xmath149 implies @xmath150 , @xmath151 and @xmath152 is the hubble parameter of de sitter space . in de sitter space in @xmath127 dimensions
the chernikov - tagirov @xcite scalar propagator reads @xcite : @xmath153+\mathcal{o}(d-4)\nonumber \,,\end{aligned}\ ] ] where we have kept the @xmath127-dimensional form of the first ( most singular ) term that may lead to singular contributions in the limit when @xmath154 .
note that @xmath155 and @xmath156 can be either @xmath157 or @xmath158 , @xmath159 , @xmath160 and @xmath161/dz$ ] as usual .
furthermore , @xmath162 is given by : @xmath163 the scalar function @xmath164 is a simple function of the geodesic distance @xmath165 in de sitter space . since we are only interested in the two wightman functions contributing to the causal propagator , we have : @xmath166 where @xmath167 .
the causal propagator ( [ commutatorpropagator ] ) thus follows as : @xmath168\right|\omega\right\rangle & = & i\delta_{-+}(x , x ' ) - i\delta_{+-}(x , x ' ) = \frac{h^{d-2}}{(4\pi)^{d/2 } } \gamma\left(\frac{d}{2}-1\right)\left\{\frac{1}{y_{-+}^{\frac{d}{2}-1 } } - \frac{1}{y_{+-}^{\frac{d}{2}-1}}\right\ } \\ & & + \frac{h^{2}}{16\pi^{2}}\sum_{n=0}^{\infty } \frac { \gamma\left(\frac{3}{2}+\nu+n\right ) \gamma\left(\frac{3}{2}-\nu+n\right ) } { \gamma\left(\frac{1}{2}+\nu\right ) \gamma\left(\frac{1}{2}-\nu\right ) } \left(\frac{y_{}}{4}\right)^{n}\left\ { \ln\left(y_{-+}\right ) - \ln\left(y_{+-}\right ) \right\}\nonumber \,.\end{aligned}\ ] ] clearly , the first term represents the standard hadamard singularity that every propagator contains .
the two terms relevant to our calculation are the ones in curly parentheses .
it can be shown that : @xmath169 + { \cal o}(d-4 ) \\
\ln\left(y_{-+}\right ) - \ln\left(y_{+-}\right ) & = & 2\pi i [ \theta(\delta\eta - r)-\theta(-\delta\eta - r ) ] \ , , \label{scalarpropagator3:ln}\end{aligned}\ ] ] where @xmath170 . while the hadamard pole results in a singular contribution at the light cone @xmath171 , the logarithmic cuts which are a mathematical description for amplification of super - hubble correlations occurring in accelerating spacetimes are responsible for finite contributions within past and future light cones .
outside past and future light cones both contributions ( [ scalarpropagator3][scalarpropagator3:ln ] ) vanish , as expected .
therefore , our causal propagator ( [ scalarpropagator2 ] ) also vanishes in these regions as it should .
although we performed the calculation in de sitter spacetime , an analogous result should hold for more general spacetimes .
now , we will show that our kernel analysis is consistent with the arguments put forward above . in the functional schrdinger picture
the causal propagator reads : @xmath172\right|\omega\right\rangle \\ & & \qquad\qquad= \int \mathcal{d}\phi\mathcal{d}\phi ' \phi(\mathbf{x } ) \phi(\mathbf{x}')\left\ { \psi^{\ast}(\phi , t_{1 } ) \psi(\phi',t_{2 } ) k(\phi , t_{1};\phi',t_{2 } ) - \psi(\phi , t_{1 } ) \psi^{\ast}(\phi',t_{2 } ) k(\phi',t_{2};\phi , t_{1})\right\}\nonumber\,.\end{aligned}\ ] ] for simplicity , we consider the free kernel ( [ propkernel ] ) . the gaussian vacuum wave functional ( see @xcite ) is given by : @xmath173 = \prod_{\mathbf{k } } \mathcal{n}_{\mathbf{k}}(t)\,\,\exp\left [ -\frac{1}{v}\big\{\phi_{\mathbf{k}}^{\ast}\,a_{k}(t)\,\phi_{\mathbf{k}}\big\ } \right ] \,,\ ] ] where @xmath174 is a normalisation constant formally given by : @xmath175\,,\ ] ] and where the @xmath176-function in fourier space is given by : @xmath177\,,\ ] ] where @xmath178 and where @xmath77 is a unit vector normal to an arbitrary plane through the origin in @xmath179-space .
we see after substitution in the functional schrdinger equation that @xmath180 and @xmath181 in fact obey ( [ motionphi2 ] ) .
this solution differs slightly when compared to @xcite .
the form of ( [ vacuumwavefunctional2 ] ) is dictated by the required invariance of @xmath182 under @xmath78 , and the difference is due to our definition of fourier decomposition in terms of complex mode functions ( [ fieldfouriertransform ] ) . in principle , we could allow for an even more general argument of the logarithm , namely @xmath183 + b_k [ \theta(\mathbf{k}\cdot\hat{\mathbf{n } } ) \phi_{\mathbf{k}}(t ) + \theta(-\mathbf{k}\cdot\hat{\mathbf{n } } ) \phi_{\mathbf{k}}^{\ast}(t)]$ ] .
then , we require @xmath184 and in order to preserve homogeneity @xmath185 . these states would correspond to the most general pure ( minimum uncertainty ) states with a `` non - zero particle number '' @xmath186 .
splitting @xmath182 into real and imaginary parts yields : @xmath187 we can in fact perform the two functional integrals in ( [ vacuumexpvalufuncschrodinger ] ) by introducing two additional sources in the exponent , varying with respect to these sources and setting them to zero subsequently : @xmath188\right|\omega\right\rangle \\ & & \quad= \left.\frac{\delta}{\delta j(\mathbf{x } ) } \frac{\delta}{\delta j'(\mathbf{x } ' ) } \int \mathcal{d}\phi\mathcal{d}\phi ' \left\ { \psi^{\ast}(\phi , t_{1 } ) \psi(\phi',t_{2 } ) k_{j , j'}(\phi , t_{1};\phi',t_{2 } ) - \psi(\phi , t_{1 } ) \psi^{\ast}(\phi',t_{2 } ) k_{j , j'}(\phi',t_{2};\phi , t_{1})\right\}\right|_{j = j'=0 } \nonumber,\end{aligned}\ ] ] where : @xmath189\,.\ ] ] after some algebra we arrive at the following intermediate result , where the two wightman functions can clearly be recognised : @xmath190\right|\omega\right\rangle & = & -\int\frac{d^{3}\mathbf{k}}{(2\pi)^{3 } } \left\{\frac{d(k , t_{1},t_{2 } ) } { 4(c(k , t_{1},t_{2})+a_{k}(t_{2 } ) ) ( b(k , t_{1},t_{2})+a_{k}^{\ast}(t_{1 } ) ) -d^{2}(k , t_{1},t_{2 } ) } \right.\\ & & + \left.\frac{d(k , t_{1},t_{2 } ) } { 4(c(k , t_{1},t_{2})-a_{k}^{\ast}(t_{2 } ) ) ( b(k , t_{1},t_{2})-a_{k}(t_{1 } ) ) -d^{2}(k , t_{1},t_{2})}\right\ } e^{i\mathbf{k}\cdot(\mathbf{x}-\mathbf{x}')}\nonumber\,.\end{aligned}\ ] ] note that both @xmath191 and @xmath192 are expressed in terms of causal green s functions and the initial state through @xmath182 .
the functional schrdinger picture shows that the evolution of correlators can be fully expressed in terms of causal propagators and initial state only .
this is to be contrasted to out - of - equilibrium field theories in heisenberg picture , where often one reads that in non - equilibrium problems there are two independent two point functions : the spectral function ( causal propagator , commutator ) and the statistical propagator ( anticommutator ) .
employing the definitions of the occurring functions in terms of the fundamental solutions in equations ( [ propkernelconst ] ) and ( [ vacuumwavefunctional2 ] ) , allows us to even further simplify this result to obtain : @xmath193\right|\omega\right\rangle & = & \hbar \int\frac{d^{3}\mathbf{k}}{(2\pi)^{3 } } g_{\mathbf{k}}(t_{1},t_{2})\mathrm{sgn}(\mathbf{k}\cdot\hat{\mathbf{n } } ) e^{i\mathbf{k}\cdot(\mathbf{x}-\mathbf{x } ' ) } \\ & = & \hbar \int\frac{d^{3}\mathbf{k}}{(2\pi)^{3 } } g_{c}(k , t_{1},t_{2})e^{i\mathbf{k}\cdot(\mathbf{x}-\mathbf{x } ' ) } \nonumber \,,\end{aligned}\ ] ] as it should be .
we have used identity ( [ causalproplinkgprop ] ) , which establishes the link with the heisenberg picture fields . note that the wightman functions in ( [ vacuumexpvalufuncschrodinger3 ] ) depend on the initial state through @xmath182 whereas the causal propagator does not .
hence , we have shown that the kernel although developed in fourier space , preserves the causal structure in terms of light cones in position space .
concluding , since the commutator is causal and the kernel is expressed solely in terms of this causal quantity , quantum mechanics in de sitter spacetime is fully causal .
this is in contrast to claims made , for example , in @xcite .
apart from @xcite , the claims in the literature @xcite are based on considering single particle quantum mechanics , in contrast to the functional schrdinger picture used in the present work .
this means that causality of quantum mechanics can be fully appreciated only within the context of ( relativistic ) quantum field theory , and can lead to misleading results when viewed within the one particle formulation of quantum mechanics .
if initial and final times are equal , the kernel ( [ propkernel][propkernelconst ] ) reduces to a delta - function , i.e. : there is no propagation of the field .
if both times differ , propagation is dictated by the commutator propagator and hence causality is preserved in general in quantum mechanics .
note finally that equations ( [ expectationvalueq2 ] ) and ( [ expectationvalueq3 ] ) support this statement .
although we have proved this consistency in de sitter spacetime explicitly , we have no reason to believe that our result does not hold in flrw and more general spacetimes .
in fact , one could take the region in spacetime where the commutator ( [ vacuumexpvalufuncschrodinger4 ] ) vanishes as the definition for light cones in general spacetimes . ] _ _
to illustrate the applicability of the kernel , we will turn our attention to three explicit examples .
we construct the kernel in both minkowski and pure de sitter spacetimes , where in the former case one can recognise the simple harmonic oscillator kernel .
finally , we will examine the evolution of the vacuum state for a non - interacting hamiltonian in general flrw spacetimes .
consider the simple harmonic oscillator toy model .
let us take the minkowski metric , i.e. : we set @xmath194 in the flrw metric . in the minimally coupled case ( @xmath195 ) , the equation of motion that the field modes @xmath180 obey follows from ( [ motionphi2 ] ) as : @xmath196 where @xmath197 . if we choose our fundamental solution of this equation of motion to be : @xmath198 then this solution obeys the correct symmetry properties . note that this choice is consistent with the wronskian normalisation condition ( [ wronskian ] ) .
again , we could have chosen an even more general solution of the form @xmath199 and require @xmath184 .
the fourier transform of the causal propagator defined in ( [ commutatorpropagatorf ] ) is hence given by : @xmath200 the kernel for the simple harmonic oscillator follows as : @xmath201 \,.\ ] ] if we let @xmath202 this is indeed in agreement with standard quantum mechanical results .
it reproduces for example @xcite where @xmath203 . as a second example we consider the kernel for the inflationary de sitter universe @xcite
the solution of ( [ motionphi2 ] ) in conformal time is thus given by : @xmath204 where @xmath205 and @xmath206 are two coefficients and : @xmath207 @xmath208 and @xmath209 are the hankel functions of the first and second kind , respectively , at order @xmath159 , see equation ( [ actions ] ) .
the causal propagator now follows as : @xmath210 \mathrm{sgn}(\mathbf{k}\cdot\hat{\mathbf{n } } ) \,.\ ] ] the causal propagator in turn fully determines our kernel ( [ propkernel ] ) . as a final example
, we can apply the kernel to an initial vacuum state for some non - interacting scalar field .
thus , initially , we start with a vacuum state @xmath4 and calculate its forward time evolution according to standard quantum mechanical lore ( [ evolution ] ) .
we employ the functional schrdinger picture as before and arrive at : @xmath211 we can now conveniently exploit equation ( [ propkernel ] ) for the kernel in the non - interacting case .
the gaussian vacuum wave functional is given by ( [ vacuumwavefunctional ] ) . in order to perform the functional integral in ( [ vacuumevolution ] )
, we switch to fourier space , complete the square and arrive at the following intermediate result : @xmath212.\ ] ] now we insert the definition of the causal propagator and the wronskian in terms of the fundamental solutions . indeed , one can show that the expression above reduces to the vacuum at time @xmath1 : @xmath213 and furthermore : @xmath214 as desired .
we constructed the kernel for a scalar field in flrw spacetimes in two cases : a free field ( [ propkernel ] ) and one coupled to a source term ( [ propkernels ] ) .
we showed that these kernels can be expressed solely in terms of the causal propagator and derivatives of the causal propagator .
we have applied the general formalism to three examples , the simple harmonic oscillator kernel ( [ shopropagatorkernelfull ] ) , the kernel in de sitter spacetime and the evolution of the gaussian vacuum wave functional ( [ vacuumevolution2 ] ) . in our analysis
the causal structure of quantum field theory in the functional schrdinger picture is manifest .
we have shown that the causal propagator , given by the vacuum expectation value of the commutator , forms the essential function in terms of which the functional kernel is expressed . therefore , our functional schrdinger picture analysis reproduces the standard ( heisenberg picture ) causality structure of flrw spacetimes .
99 a. h. guth and s. y. pi , the quantum mechanics of the scalar field in the new inflationary universe , phys .
d * 32 * ( 1985 ) 1899 .
t. prokopec and g. i. rigopoulos , decoherence from isocurvature perturbations in inflation , jcap * 0711 * ( 2007 ) 029 [ arxiv : astro - ph/0612067 ] .
j. f. koksma , decoherence of cosmological perturbations , m.sc . thesis ( 2007 ) , + http://www1.phys.uu.nl/wwwitf/teaching/thesis.htm .
s. weinberg , quantum contributions to cosmological correlations , phys .
d * 72 * ( 2005 ) 043514 [ arxiv : hep - th/0506236 ] .
t. prokopec and e. puchwein , photon mass generation during inflation : de sitter invariant case , jcap * 0404 * , 007 ( 2004 ) [ arxiv : astro - ph/0312274 ] .
m. e. peskin and d. v. schroeder , an introduction to quantum field theory , westview press ( 1995 ) .
r. jackiw , analysis on infinite dimensional manifolds : schroedinger representation for quantized fields , presented at seminar on higher mathematics , montreal , canada , jun 1988 .
r. jackiw , functional representations for quantized fields , published in `` 1st asia pacific workshop on high energy physics , '' sinagpore , jun 21 , 1987 and 6th symp . on theoretical physics , seoul , korea ,
jul 1987 and mathematical quantum field theory , montreal , canada , sep 1987 .
j. guven , b. lieberman and c. t. hill , schrdinger picture field theory in robertson - walker flat spacetimes , phys .
d * 39 * , 438 ( 1989 ) .
t. prokopec and g. rigopoulos , in progress .
n. a. chernikov and e. a. tagirov , quantum theory of scalar fields in de sitter space - time , annales poincare phys .
theor . a * 9 * ( 1968 ) 109 .
b. garbrecht and t. prokopec , fermion mass generation in de sitter space , phys .
d * 73 * , 064036 ( 2006 ) [ arxiv : gr - qc/0602011 ] .
d. v. long and g. m. shore , the schrdinger wave functional and vacuum states in curved spacetime , nucl .
b * 530 * , 247 ( 1998 ) [ arxiv : hep - th/9605004 ] .
j. berges , introduction to nonequilibrium quantum field theory , aip conf .
* 739 * ( 2005 ) 3 [ arxiv : hep - ph/0409233 ] . k. nomoto and r. fukuda , causality in the schrdinger picture : diagrammatical approach , phys .
d * 46 * , 1680 ( 1992 ) .
g. c. hegerfeldt , remark on causality and particle localization , phys .
d * 10 * ( 1974 ) 3320 .
g. c. hegerfeldt , violation of causality in relativistic quantum theory ?
* 54 * ( 1985 ) 2395 .
g. c. hegerfeldt , instantaneous spreading and einstein causality in quantum theory , annalen phys .
* 7 * ( 1998 ) 716 [ arxiv : quant - ph/9809030 ] , and references therein .
h. halvorson and r. clifton , no place for particles in relativistic quantum theories ?
* 69 * ( 2002 )
1 [ arxiv : quant - ph/0103041 ] . h. halvorson and m. muger , algebraic quantum field theory , [ arxiv : math - ph/0602036 ] .
j. fernando perez and i. f. wilde , localization and causality in relativistic quantum mechanics , phys .
d * 16 * ( 1977 ) 315 .
j. j. sakurai , modern quantum mechanics , addison wesley longman ( 1994 ) .
|
we construct the quantum mechanical evolution operator in the functional schrdinger picture for a scalar field in spatially homogeneous flrw spacetimes when the field is a ) free and b ) coupled to a spacetime dependent source term . the essential element in the construction
is the causal propagator , linked to the commutator of two heisenberg picture scalar fields .
we show that the kernels can be expressed solely in terms of the causal propagator and derivatives of the causal propagator .
furthermore , we show that our kernel reveals the standard light cone structure in flrw spacetimes .
we finally apply the result to minkowski spacetime , to de sitter spacetime and calculate the forward time evolution of the vacuum in a general flrw spacetime .
|
deviations from thermodynamic equilibrium are of fundamental importance in the early universe .
the examples are plentiful , for example the freeze - out of photon thermodynamic equilibrium due to hydrogen recombination @xcite or the equivalent freeze - out of neutrinos slightly prior to electron - positron annihilation @xcite .
these phenomena have been investigated many times in the literature , but almost always in the context of deviations from chemical equilibrium , that is , kinetic or scattering equilibrium is assumed to hold at all times . in the limit of boltzmann statistics the single particle distribution then takes the form @xmath0 where @xmath1 is a pseudo - chemical potential , independent of @xmath2 . in recent years
there has been some discussions as to the validity of this assumption , especially concerned with the decoupling of neutrinos in the early universe @xcite . in this case
it has been found by several authors that the effects are small , but not negligible . in the present paper
we will discuss chemical versus scattering equilibrium in a general context , where we limit ourselves to generic expressions for the matrix elements .
dolgov @xcite has provided a general discussion of deviations from kinetic equilibrium caused by the overall expansion of the universe , looking only at elastic scattering .
the case we treat is where both scattering and annihilation is taken into account to solve the full boltzmann equation , however we resort to a numerical solution of the boltzmann equation . in general , for dirac fermions it can be assumed that s - wave ( @xmath3 ) annihilation is by far the most important for non - relativistic particles .
the reason is the following : since particle and antiparticle are two different particles there are no anti - symmetry requirements and the annihilation can proceed via s - waves . since higher order terms
@xmath4 are suppressed by factors of @xmath5 @xcite , where @xmath6 is the relative velocity of the incoming particles , s - wave annihilation is dominant . for majorana particles particle and antiparticle
are identical objects and the combined wave function must be antisymmetric . therefore annihilation proceeds primarily via p - waves ( @xmath7 ) , meaning that the annihilation amplitude is suppressed relative to the dirac case . in the case of s - wave annihilation
the matrix element is not momentum dependent and the annihilation amplitude is constant . for p - wave annihilation there
will , as mentioned , be a dependence on the square of the relative velocity of the two incoming particles . for non - relativistic particles we can therefore in general write the annihilation amplitude as @xmath8 where @xmath9 is a dimensionless coupling constant .
we shall use this expression for the matrix element even for relativistic decoupling , although that is of course at best only an approximation to the true matrix element .
for example , in the case of massive dirac neutrinos , the matrix element will be strongly energy dependent as long as the neutrinos are relativistic , but approach a constant as the neutrinos become non - relativistic @xcite .
however , taking our approach , although quantitatively not accurate for relativistic decoupling , still gives a very clear picture of when non - equilibrium effects are important .
the elastic scattering part will , in general , have a momentum dependent amplitude .
this is for example the case for scattering of heavy neutrinos on massless particles via the weak interaction . here
, the squared matrix element is proportional to the squared energy of the scatterer . for the sake of simplicity we shall deal only with three cases , the first being no elastic scattering at all and the second elastic scattering with a constant amplitude .
the first case will represent an extreme upper limit to the size of non - equilibrium effects .
the second case , albeit not realistic , serves to indicate how elastic scattering influences the shape of the massive particle spectra .
the third case we shall treat is the case where full scattering equilibrium is maintained at all times .
when discussing non - equilibrium effects in the early universe the relevant equation is the boltzmann collision equation .
since we are not discussing such issues as medium effects we shall use the single - particle boltzmann equation . also , we restrict the calculation to using boltzmann statistics , which simplifies the calculations while leaving the essential physics intact
. then the boltzmann equation takes the form @xcite @xmath10 where @xmath11 @xmath12 being the hubble parameter @xmath13 . on the right - hand side ,
@xmath14 and @xmath15 represent annihilations and elastic scatterings respectively .
these collision terms can be written generically as @xmath16 & = & \frac{1}{2e_{1}}\int d^{3}\tilde{p}_{2 } d^{3}\tilde{p}_{3}d^{3}\tilde{p}_{4 } ( f_{3}f_{4}-f_{1}f_{2 } ) \label{integral}\\ & & \,\,\ , \times s \sum | \ ! m \ ! |^{2}_{12\rightarrow 34}\delta^{4 } ( { \it p}_{1}+{\it p}_{2}-{\it p}_{3}-{\it p}_{4})(2\pi)^{4 } , \nonumber\end{aligned}\ ] ] as long as we are only concerned with 2-body collisions . here
we have @xmath17 .
@xmath18 is a symmetrisation factor of 1/2 ! for each pair of identical particles in initial or final states @xcite , and
@xmath19 is the interaction matrix element squared and spin - summed .
@xmath20 is the four - momentum of particle @xmath21 .
now , the next issue is to actually perform the relevant collision integrals .
it is in general possible to integrate @xmath14 and @xmath15 down to two dimensions , depending on the actual shape of @xmath22 .
the details of these calculations are given in the appendix .
the usual approach to solving the above equation , eq .
( [ eq : boltz ] ) , is to assume full scattering equilibrium ( our third case ) .
then the elastic scattering term is identically zero at all times , and the boltzmann can easily be integrated to yield @xcite @xmath23 where @xmath24 is the actual number density and @xmath25 is the equilibrium number density .
@xmath26 is the velocity - averaged annihilation cross - section @xcite . as for the general expansion of the universe we shall assume that the universe is completely radiation dominated and further we will ignore entropy production in order to make the discussion completely generic .
including the correct expansion law by using the friedmann equation and the equation of energy conservation will only perturb the general results slightly in most cases . using the friedmann equation @xcite ,
@xmath27 one then arrives at the following time - temperature relation @xmath28 where @xmath29 , and @xmath30 is the total radiation energy density .
in order to investigate the non - equilibrium effects in a quantitative way we still have to calculate the remaining two dimensions of the collision integrals , eq .
( [ integral ] ) , numerically .
we have done this in order to estimate whether non - equilibrium effects can ever be important during particle freeze - out .
notice that we do not make any assumptions as to the size of the non - equilibrium effects .
that is , we do not assume _ a priori _ that they are small , as was done for instance in ref .
@xcite , but instead we perform a full numerical integration of the whole system . in all calculations
we will assume that the massless particles are in full thermodynamic equilibrium because of their interactions with other light particles , an assumption which simplifies the calculations .
for both dirac and majorana particles one can expect that the coupling constant will in general be the same for annihilation and scattering , e.g. @xmath31 for weak interactions ffor a non - relativistic dirac neutrino , @xmath32 , assuming only the annihilation channel @xmath33 ) or @xmath34 for electromagnetic interactions .
we can therefore scale the amplitude in units of a single coupling constant , @xmath9 , so that @xmath35 = 11truecm clearly , the size of the non - equilibrium effects will depend on @xmath9 .
if @xmath9 is very small , particles will decouple while still relativistic .
this of course would not produce any non - equilibrium effects at all . on the other hand ,
if the particles are extremely non - relativistic before they decouple , the effects are also likely to be small for the following reason .
for very non - relativistic particles in equilibrium , the number density is suppressed exponentially relative to the massless species .
therefore the annihilation rate is also suppressed exponentially , while the scattering rate is suppressed at most by some power of @xmath36 .
this will increase the importance of elastic scattering relative to annihilation and lessen any non - equilibrium effects , even though the scattering on massless particles is essentially momentum conserving . _
a priori _ one therefore expects that non - equilibrium effects are the largest for semi - relativistic decoupling . in fig . 1 we have plotted the final number density of massive particles in comoving coordinates as a function of a dimensionless interaction strength , defined as @xmath37 there is a pronounced difference here between the two cases of @xmath3 and @xmath7 . for strong interactions ( large @xmath38 )
, the final number density is much higher for the velocity dependent interaction because of the large suppression factor @xmath39 . on the other hand , for relativistic decoupling , the velocity dependent interaction is stronger by the same factor for the same value of @xmath38 and therefore the final number density in this case is lower than for the constant matrix element .
also , the freeze - out process proceeds much faster in the @xmath7 case than in the @xmath3 because the annihilation amplitude falls off with decreasing temperature .
= 8truecm [ fig2 ] in fig .
2 we show how the annihilation actually proceeds in the two cases .
this example is taken for @xmath40 .
clearly , for the constant matrix - element decoupling is very prolonged , taking place over a wide range in @xmath41 . for the velocity dependent matrix element ,
decoupling is much more sudden , essentially completing over a factor of two in @xmath41 .
this difference will of course give rise to very different non - equilibrium effects .
3 shows the deviation in final number density between relative to that found by using the integrated boltzmann equation .
for the solution of the full equation ( the full lines ) we see the expected trend , namely that relativistic decoupling leads to no significant non - equilibrium effects and the same being true for very non - relativistic decoupling .
the effect has a maximum for semi - relativistic decoupling , where the difference between using the full boltzmann equation and the integrated version can be as large as 10 - 15% .
however , even for strongly non - relativistic decoupling there still is a non - negligible effect because of residual annihilations taking place after decoupling from both scattering and chemical equilibrium .
thus , one can not entirely neglect non - equilibrium effects when doing this sort of calculation . indeed , for the special case of massive neutrino decoupling it has been demonstrated previously in the literature that out - of - equilibrium effects can be significant @xcite .
note also the interesting feature that for the case of a constant matrix element , the annihilation term will itself preserve scattering equilibrium if the massive particles are non - relativistic .
this can be seen directly from the kernel in the annihilation integral , eq .
( [ eq : kernel ] ) . the dotted curve in the upper panel of fig .
3 shows an example of this behaviour . here
, the size of the non - equilibrium effects also drops for strong interactions because complete equilibrium is maintained until the particles are strongly non - relativistic , where after @xmath14 preserves scattering equilibrium independent of the interaction strength . for the velocity dependent matrix element
the situation is the reverse because the annihilation integral kernel , eq .
( [ eq : kernel2 ] ) , does not preserve kinetic equilibrium . rather , high momentum states annihilate much faster than low momentum ones , so that kinetic equilibrium is quickly destroyed .
this can lead to very large errors in the computed number density if one uses the approach of only taking into account the annihilation term in the collision integral .
= 11truecm the sign of the non - equilibrium effect is also opposite for the @xmath3 and @xmath7 scenarios . for the constant matrix element
low momentum states will , in general , annihilate faster than high momentum ones so that maintaining kinetic equilibrium leads to a lower annihilation rate . for the velocity dependent matrix element high momentum states deplete quickly so that annihilation is slowed unless they are refilled by scattering .
this effect is known from the case of massive majorana neutrinos @xcite . finally , in fig .
4 we show an example of the actual distribution functions obtained by different methods , taken as a snapshot at @xmath42 for @xmath40 . what is shown is the ratio between the actual distribution function and the corresponding equilibrium distribution with the same number density .
obviously , the distribution one obtains by using the integrated equation is an equilibrium distribution per default .
however there can be huge differences between that and the actual distribution function obtained from the full boltzmann equation .
also , there is a large difference between the full solution and the one obtained by using only the annihilation term .
our results clearly show that for the case of semi - relativistic decoupling one can not safely ignore the non - equilibrium effects pertaining to the deviations from scattering equilibrium .
this will for instance be the case with mev mass neutrinos which can influence big bang nucleosynthesis . however , for either relativistic or non - relativistic decoupling the effects are negligible .
thus , for most cold dark matter candidates which per definition decouple while strongly non - relativistic , , see for instance ref .
@xcite ] it is completely safe to assume full scattering equilibrium until long after decoupling from chemical equilibrium . as a simple example we shall consider a massive dirac neutrino which is thermally produced in the early universe
this particle is an excellent cdm candidate if it has a mass in the 1 gev region , around the lee - weinberg limit @xcite .
the dimensionless interaction strength for this particle is given by @xmath43 assuming that the only relevant annihilation process is @xmath44 . in fig .
5 we have plotted @xmath38 as a function of the heavy neutrino mass . for a mass of 1 gev , @xmath45 , meaning that there are essentially no deviations from kinetic equilibrium . as a note of caution
we stress that we have assumed a scattering matrix element of the same magnitude as the annihilation matrix element which is incorrect .
rather , the scattering matrix element is @xmath46 which is lower by a factor of @xmath47 .
even so there will be no discernible effect due to deviations from scattering equilibrium for such a heavy neutrino .
the same will be true for super - symmetric dark matter particles which are likely to have a mass in the multi - gev region or above @xcite . however , there are other examples where non - equilibrium effects are sizeable , for example the previously mentioned example of an mev neutrino .
= 8truecm it was also shown that taking into account only the annihilation term in the full boltzmann equation can lead to large errors in the final number density if the matrix element is velocity dependent .
this was for instance the case in the work by kawasaki et al .
@xcite .
finally , in the present paper we have not discussed possible deviations from equilibrium in the massless particle species induced by interactions with the massive particles .
this effect is for instance of importance when the massless particles themselves are decoupling from equilibrium at the same temperature as the massive ones .
an example is the decoupling of mev @xmath48-neutrinos where the induced deviations from equilibrium in the electron neutrino distribution causes big bang nucleosynthesis to change @xcite .
this work was supported by the theoretical astrophysics center under the danish national research foundation .
this appendix deals with the phase - space integrations of the collision operator of eq .
( [ integral ] ) .
the principal method of integration was developed in ref .
@xcite and in the following we only give a brief outline and further develop the integrations in the non - relativistic case . the collision integral in the general two - body collision integral can be reduced to two dimensions .
let the process be @xmath49 then the result is @xmath50 = \int dp_2 dp_3 [ j(p_1,p_2,p_3)]^{\inf[1,\alpha^+]}_{\sup[-1,\alpha^- ] } ( f_3 f_4 - f_1 f_2),\ ] ] where @xmath51 the function @xmath52 is the matrix element integrated over all variables except @xmath53 and @xmath54 @xcite . in the non - relativistic case
the expression for @xmath55 simplifies further , and it becomes possible to do one more integration . in this case on gets @xmath56 in the case where the light particles are in equilibrium we can further integrate over either @xmath57 ( annihilation ) or @xmath58 ( scattering ) .
we shall treat the case of annihilation first . for the case of a constant matrix element one
gets the result @xmath59^{\inf[1,\alpha^+]}_{\sup[-1,\alpha^- ] } ( f_3 f_4 - f_1 f_2 ) & = & g^2 \frac{1}{2 } f_1 ( p_1,p_2 ) , \label{eq : kernel}\end{aligned}\ ] ] where @xmath60 for a matrix element that is proportional to @xmath61 , one gets instead @xmath62^{\inf[1,\alpha^+]}_{\sup[-1,\alpha^- ] } ( f_3 f_4 - f_1 f_2 ) = \frac{p_1 ^ 2+p_2 ^ 2}{m^2 } i. \label{eq : kernel2}\ ] ] using these expressions we have reduced the pair - annihilation integral to a one - dimensional integral which is quite straightforward to calculate numerically , @xmath63 also , it is easy to see that the velocity dependent matrix element is suppressed relative to the constant one by a factor @xmath64 , which is large in the non - relativistic case . from the above equations one can also easily calculate the velocity averaged annihilation cross sections for the two cases
if one assumes scattering equilibrium at all times @xmath65 these can be plugged into the standard integrated boltzmann equation @xmath66 finally , the evaluation of the scattering integrals in the non - relativistic limit is also quite straightforward . here , the integral must be over @xmath58 to avoid integrating over the distribution function of the heavy particle .
the result of this integration is @xmath67^{\inf[1,\alpha^+]}_{\sup[-1,\alpha^- ] } ( f_3 f_4 - f_1 f_2 ) & = & g^2 \frac{2}{p_1 p_3}\left[\exp\left(-\frac{|p_1-p_3|}{2}\right ) \right . \nonumber \\ & & \left .
-\exp\left(-\frac{p_1+p_3}{2}\right)\right ] f_2 ( p_1,p_3),\end{aligned}\ ] ] where @xmath68 altogether then we can write the elastic scattering term as @xmath69 for a good review , see for instance p. j. e. peebles , principles of physical cosmology , princeton university press ( 1993 ) .
s. dodelson and m. s. turner , phys .
d * 46 * ( 1992 ) 3372 ; a. d. dolgov and m. fukugita , phys .
d * 46 * ( 1992 ) 5378 ; s. hannestad and j. madsen , phys . rev .
d * 52 * ( 1995 ) 1764 ; a. d. dolgov , s. h. hansen and d. v. semikoz , nucl .
* b503 * ( 1997 ) 426 .
s. hannestad and j. madsen , phys .
* 76 * ( 1996 ) 2848 , erratum _ ibid . _
* 77 * ( 1996 ) 5148 ; s. hannestad and j. madsen , phys .
d * 54 * ( 1996 ) 7894 .
a. d. dolgov , s. h. hansen and d. v. semikoz , hep - ph/9712284 ( 1997 ) ( submitted to nucl .
m. kawasaki , g. steigman , and h .- s .
kang , nucl . phys .
* b403 * ( 1993 ) 671 .
a. d. dolgov , nucl . phys . *
b496 * ( 1997 ) 437 .
a. d. dolgov , s. pastor and j. w. f. walle , phys . lett .
* b383 * ( 1996 ) 193 . k. kainulainen , talk given at the 17th international conference on neutrino physics and astrophysics ( neutrino 96 ) , in _
neutrino 96 _ , eds .
k. enqvist , k. huitu and j. maalampi , world scientific ( 1996 ) , pp .
461 - 466 .
b. d. fields , k. kainulainen and k. a. olive , astropart .
* 6 * ( 1997 ) 169 .
k. griest and m. kamionkowski , phys .
* 64 * ( 1990 ) 615 .
j. bernstein , kinetic theory in the expanding universe , cambridge university press ( 1988 ) .
r. v. wagoner , in physical cosmology - les houches ( 1979 ) , eds .
r. balian , j. audouze and d. n. schramm .
e. w. kolb and m. s. turner , the early universe , addison wesley ( 1990 ) .
p. gondolo and g. gelmini , nucl .
( 1991 ) 145 .
b. w. lee and s. weinberg , phys .
( 1977 ) 165 ; m. i. vysotskii , a. d. dolgov and ya . b. zeldovich , jetp lett .
* 26 * ( 1977 ) 188 ; e. w. kolb and k. a. olive , phys . rev .
d * 33 * ( 1986 ) 1202 , erratum _ ibid . _
* 34 * ( 1986 ) 2531 . for a recent review ,
see n. fornengo , talk given at dark matter , trieste ( 1997 ) , astro - ph/9804295 .
s. hannestad and j. madsen in ref .
|
we investigate the possible effects that deviations from kinetic equilibrium can have on massive particles as they decouple from chemical equilibrium .
different methods of solving the boltzmann equation yield significantly different relic number densities of such particles .
general considerations concerning the dirac or majorana structure of the particles are discussed .
it is shown that non - equilibrium effects are small for particles decoupling while strongly non - relativistic , as will be the case for most cold dark matter candidates .
+ pacs : 95.35.+d , 98.80.cq , 95.30.cq , 14.60.st + keywords : early universe , dark matter , neutrino physics
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.